Solutions Manual

to problems in

Failure, Fracture, Fatigue An Introduction

by Tore Dahlberg Anders Ekberg

Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

This manual contains solutions to problems in the textbook

Failure, Fracture, Fatigue - An Introduction Studentlitteratur, Lund 2002, ISBN 91-44-02096-1

At present, solutions to all problems given in Chapters 1 to 6 and Chapters 8 and 9 are available in this document (Chapter 7 does not contain any problems and solutions to the problems in Chapter 10 will be available later). It is our hope that these solutions will be to some help for teachers and students when studying the topics covered in the textbook. Samples of examinations are available (2010) on the home page of Solid Mechanics, IEI, at Linkping University, Sweden, address: www.solid.iei.liu.se under the heading "Examinations" and course "Damage mechanics and life analysis TMHL61". Linkping and Gothenburg in February 2010 The authors

Address to Tore Dahlberg after May 2010: smark 6, 563 92 Grnna, Sweden
Page 1:2

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 1
Introduction, failure mechanisms
Problems with solutions Elastic deformations 1/1.
Side view : Top view :
P

Cross section: B t
H H >> t B >> t

A cantilever beam, length L, carries a force P at its free end. The beam cross section is an ideal I profile, i.e., the area of the flanges is width B by height t each (B >> t), and the area of the web can be neglected. The height is H, where H >> t, see figure. When the beam is loaded by the force P, the deflection 1 becomes
P L3 H 2 Bt H 2 1 = where I = 2 Bt = 3 EI 2 2

Thus, 1 = 2P L 3 / 3E BtH 2 = 0.67P L 3 / E BtH 2.

Side view :
L

Top view :
x

Cross section : 2 B (1 - x / L) t
H

It has been found that this deflection is too large. Therefore, without increasing the weight (the mass) of the beam, another form of the cantilever beam cross section is tried. Material P is moved from the flanges at the free end and placed at the flanges at the fixed end to give a beam cross section (still an ideal I profile) with a width varying with the coordinate x, see figure. The height of the profile is still H, but H >> t the width of the flanges is now B(x) = B >> t 2B(1 x /L). What deflection 2 will now be obtained when the load is applied at the free end?
P
Chapter 1 Page 1:1

SOLCHAP1.DOC/2010-01-29/TD

Solution: Use the differential equation

EI (x ) w (x ) = M (x ) (a) to determine the deflection w(x) of the beam. The bending moment M(x) in the beam becomes M(x) = PL (1 x /L), and the second moment of area becomes x x H2 I = 2 2B 1 t = Bt H 2 1 L 2 L (b)

The direction of the moment M(x) has been selected to give a positive deflection w(x) downwards. Entering M(x) and I(x) in the differential equation gives PL M (x ) PL (1 x /L ) = w (x ) = = EI (x ) E Bt H 2(1 x /L ) E Bt H 2 Integration gives
w (x ) = PL x + C1 E Bt H 2 (d) (e)

(c)

and

PL x 2 w (x ) = + C1 x + C2 E Bt H 2 2

Boundary conditions at x = 0 give (notation BC for boundary condition) BC 1: w (0) = 0 gives C2 = 0

BC 2: w (0) = 0 gives C1 = 0

(f) (g)

The deflection w(x) then becomes

PL x 2 w (x ) = E Bt H 2 2 (h)

and at the free end (at the force P) one obtains

PL L 2 PL 3 2 = w (L ) = = E Bt H 2 2 2E Bt H 2 (i)

It is concluded that the rearrangement of the width of the flanges makes the deflection at the free end to decrease from 0.67 PL3/EBtH2 to 0.5 PL3/EBtH2, i.e., the deflection decreases by 25 per cent. Answer: The rearrangement of the width of the flanges causes the deflection at the free end to decrease from 0.67 PL3/EBtH2 to 0.5 PL3/EBtH2.
Page 1:2 Chapter 1

Yielding 1/2. The stress components in a part of a structure have been calculated to x = 120 MPa, y = 80 MPa, and xy = 60 MPa (all other stress components are zero). For this structure a safety factor s = 1.5 with respect to yielding is required. What yield strength Y should the material have to fulfil this condition? Investigate both the Tresca and the von Mises yield criteria. Solution: First, determine the principal stresses in the material. In the xy-plane one obtains
x y 2 2 + xy = 100 63.2 (MPa) 2 and do not forget the third principal stress: 3 = 0. x + y 1, 2 = 2

(a)

Thus, the largest principal stress is 1 = 163.2 MPa and the smallest principal stress is 3 = 0. The effective stress according to Tresca then is eT = 1 3 = 163.2 MPa. Using safety factor s = 1.5 on the yield limit gives Y 163.2 = (a) which gives Y = 245 MPa 1.5 The von Mises equivalent stress is
2 2 vM 2 x + y x y + 3xy = 148.3 MPa e =

(b)

Again, using safety factor s = 1.5 on the yield limit gives Y 148.3 = which gives Y = 222 MPa 1.5

(c)

Thus, according to the more conservative criterion (the Tresca criterion) the yield strength of the material should be Y = 245 MPa. Answer: The yield strength of the material should be, according to the Tresca criterion, Y = 245 MPa (and the von Mises criterion gives Y = 222 MPa).

Chapter 1

Page 1:3

Yielding and plastic collapse 1/3.

P h b stress

Material:

strain

A cantilever beam of length L and with a rectangular cross section (base b, height h) is loaded by the force P at the free end. (a) Determine the load P = Pelast the beam may be loaded with without inducing any plastic deformation in the beam (i.e., Pelast is the maximum load not giving any yielding in the beam).

(b) Determine the maximum force Pmax the beam may be loaded with. Assume that the material is linearly elastic, ideally plastic with yield limit Y. (c) Finally, determine the ratio of the load Pmax to load Pelast. Solution: (a) To start with, the elastic solution giving Pelast will be determined. The stress in the beam is obtained as Mb = (a) Wb where Mb = Pelast L, and Wb is the section modulus: Wb = bh2/6. When = Y one obtains
Mb Y b h 2 = Pelast = L 6L (b)

x z
Y

(b) The maximum force that may be applied to the beam will produce a plastic hinge at the support of the beam. The bending moment Mf needed to produce the hinge is
0 h /2

Mf = z dA = Y z b dz + h /2 0

Y b h 2 Y z b dz = 4

(c)

Thus Pmax = Mf /L = Ybh2/4L. (c) The ratio Pmax /Pelast becomes Pmax /Pelast = 3/2. Answer: (a) Pelast = Y bh 2 / 6L, (b) Pmax = Y bh 2 / 4L, and (c) Pmax /Pelast = 3/2.

Page 1:4

Chapter 1

Plastic collapse 1/4.

P L /3 z x

2 L /3 B

The beam AC is rigidly supported (clamped) at end A and simply supported at end C. The beam has length L and a rectangular cross section with base b and height h, and it is loaded at x = L /3 by a force P.

(a) Determine the force P = P0 giving one plastic hinge on the beam (to give a plastic hinge, a cross section must be fully plastic). (b) Determine the force P = Pf giving collapse of the structure. (For collapse, two plastic hinges are needed. Here they will appear at A and B.) The material is linear elastic, ideally plastic with yield limit Y. Solution:
A
MA L /3 P

2 L /3

- 5 PL /27

8 PL /81

First determine the distribution of the bending moment M along the beam when the beam is C fully elastic. Handbook formulae give, for an elastic beam, the moment MA at support A as RC MA = 5PL /27. The reaction force at support C becomes RC = P /3 5P /27 = 4P /27. The x bending moment MB at B becomes MB = RC 2L /3 = 8PL /81. Note that MA and MB have different directions, see figure.

According to the conventions used here the moment MA is negative, but in most cases only the numerical value of the moment (and not the direction) is of interest. Therefore, only numerical values of the moments will be discussed below. Note also that, numerically, MA > MB, implying that plastic yielding will start at point A. Next, determine the bending moment M = Mf needed to create a plastic hinge on the beam. When a plastic hinge appears, half the cross-sectional area A (A = b h) is yielding in tension and half of the area A is yielding in compression. This gives (see Problem 1/3)
Mf = x z dA = 2 A 0
h /2

bh 2 1 h2 Y z b dz = 2 Y b = Y 2 2 4

(a)

Chapter 1

Page 1:5

(a) Question (a) can now be answered. The first plastic hinge will appear in the beam when the largest moment (here MA) reaches the value Mf. (It is assumed that MA > MB is valid also during yielding at A.) Thus, MA = Mf gives (numerically)
bh 2 5PL Mf = Y = 4 27 27 bh 2 which gives P = P0 = 20 L Y (b,c)

Thus, at load P = P0 = 1.35 Y bh2/L one plastic hinge has developed (at A) in the beam.
P = P0

2 L /3 MB
RC P > P0

A L /3

2 L /3

C
RC

MA= Mf plastic hinge MA = Mf A

P = Pf

plastic hinges Collapse mode P = Pf

During the loading of the beam, after yielding has started at point A, the load distribution in the structure may change. This implies that the ratio MA /MB of the bending moments may change during yielding at A. Therefore, confirm that MA is larger than MB also when the hinge appears at A, i.e., confirm that MA = Mf > MB when P = P0. When MA = Mf and P = P0, the support reaction RC becomes RC = P0 /3 Mf /L, and the moment MB at B becomes MB = 2RC L /3 = 2Y bh2/15. This moment is less that Mf, which implies that the beam is still elastic at B when the hinge appears at A, i.e., at least part of the cross section at B is elastic, because the moment at MB is not large enough to create a plastic hinge there.

(b) The force P can be further increased, beyond P0, because the part AC of the beam is still elastic. When MA = Mf the beam AC acts as a simply supported beam, supported at A and C, and loaded with a moment Mf at A and a force P (> P0) at B. Collapse of the beam will appear when the bending moment at B reaches Mf, i.e., when a plastic hinge is obtained also at B. When the load P > P0 has increased so much that it produces a hinge also at B, thus giving MB = Mf, the structure will collapse. The force P = Pf that gives a hinge also at B can be determined in the following way: the bending moment

Page 1:6

Chapter 1

at B is MB = Mf = RCf (2L /3), which gives RC = RCf = MB /(2L /3) = 3Mf /2L, see figure below. The loading of the full beam AC then is MA = Mf at A, P = Pf at B, and RC = RCf = 3Mf /2L at C, see figure. The force P = Pf is still unknown.
Pf

Mf

2 L /3
R C f = 3 Mf / 2 L

A L /3 Mf B

Pf

2 L /3

Moment equilibrium with respect to point A gives an equation containing the unknown force P = Pf. One obtains L L 3M f MA = Mf = Pf RCf L = Pf (d) 3 3 2 from which is solved

3 15 bh 2 5 Pf = Mf = Y (e) R C f = 3 Mf / 2 L 2 L 8 L Thus, after the first hinge has appeared at load P = P0 = 1.35 Y bh2/L, giving one plastic hinge at A, the force P may be increased to P = Pf = 1.875 Y bh2/L, at which load two plastic hinges are obtained, and the structure will collapse. Answer: P0 = 1.35 Y bh2/L gives one plastic hinge at A, and Pf = 1.875 Y bh2/L gives plastic collapse (two hinges).

Chapter 1

Page 1:7

Plastic instability 1/5.

A0 , L0 A, L

Determine the maximum tensional force Pmax the bar in the figure (see also Section 1.4.4) may be subjected to. Assume that the material is hardening due to plastic deformation and use the stress-strain relationship
= 0 m

Material: stress
strain

Use true stress and natural (logarithmic) strain. The true stress is defined as P = A

where A is the (contracted) cross-sectional area obtained after loading, i.e. A < A0 when P > 0, and A depends on P. The logarithmic strain is defined as
= ln L L0

where L is the length of the bar when loaded, i.e. L > L0 when P > 0, and L depends on P. Also, remember that plastic deformation will take place with no change of volume. Solution: As in the example on plastic instability, see Section 1.4.4 in the textbook, one has equilibrium: P = A (a) the volume is constant: large strain:
V = A0 L 0 = A L = ln L L0 (b) (c) (d) (e) (f)

Differentiate (a), (b) and (c). It gives dP = dA + A d

dV = A dL + L dA = 0 d = 1 dL dL = L /L0 L0 L

Plastic instability occurs when the load cannot be increased any more, i.e.
Page 1:8 Chapter 1

when dP = 0. It gives
dA + A d = 0 A dL + L dA = 0 dL = L d (g) (h) (i) (j) (k)

Eliminate A and L from (g), (h) and (i). It gives A L d d = d = d = ; thus = dA dL d d Introduce the material relationship = 0 m into (j). It gives
0 m = 0 m m 1 giving = m

Thus, the loading force P reaches its maximum value P = Pmax when = m. Then L0 Pmax = A = 0 m m A = 0 m m A0 (l) L The ratio L0 /L is obtained from equation (c). By use of = m, one obtains L0 L L giving em = , or, (m) = e m = m = ln L0 L0 L Finally, entering (m) into (l) gives
Pmax = 0 m m A0 e m (n)

Alternative solution: The loading force P may be written, by use of (a) and (c), L0 P = A = 0 m A0 = 0 m A0 e L Determine the maximum value of P. One has dP = 0 A0 {m ( 1) e + m m 1 e } = 0 d from which = m is obtained. Entering = m into (o) gives
P = Pmax = 0 m m A0 e m

(o)

(p) (q)

(Verify that a maximum value, and not a minimum value, has been determined.) Answer: Plastic instability will occur when Pmax = 0 m m e m A 0.

Chapter 1

Page 1:9

Stationary creep 1/6.

h h p r

p
i o

A tube subjected to an internal pressure p is made of two circular cylinders of different materials. The outer cylinder has inner radius r, wall thickness h (h << r), and is made of a Hooke material with modulus of elasticity Eo = 207 GPa (i.e. linear elastic material with constitutive equation = /Eo). The inner cylinder has outer radius r, wall thickness h, and is made of a Norton material. Thus, before the loading has been applied the inner cylinder fits exactly in the outer one.

The material equation of the inner cylinder reads | | n sgn = + Ei n t * where Ei is the elastic modulus of the material of the inner cylinder (Ei = 69 GPa), n and n are material parameters (n > 1), sgn means sign of and is +1 if is positive and 1 if is negative, and t* is a reference time. (a) Calculate the stresses in the two cylinders immediately after the pressure p has been applied to the inner cylinder (no longitudinal stresses will appear, only circumferential). (b) Investigate how the stresses will change with time in the two cylinders. Solution: (a) When, at time t = 0, the pressure p is applied to the inner cylinder, a contact pressure q will arise between the inner and the outer cylinder. The circumferential stress i in the inner cylinder becomes i = (p q) r /h and in the outer cylinder the circumferential stress o becomes o = qr /h (index i for the inner cylinder and index o for the outer). The contact pressure q is unknown, see figure.

q
o o

q p
i i

Page 1:10

Chapter 1

The contact pressure q is obtained from the deformation condition that the two cylinders have the same radial expansion u. Thus, ui = uo. For circular symmetry the circumferential strain equals = u /r (see equation (1.18b) in Section 1.2.6 in the textbook) where u is the radial displacement (radial expansion) of the thin-walled cylinder. One obtains ui = r i = r o = uo (a) This gives, at time t = 0,
i o = Ei Eo (b)

Entering i = (p q) r /h and o = qr /h in (b) gives Ei p q =q Eo Using the numerical values of Ei and Eo one obtains q = 3p /4 (at t = 0).

(c)

Thus, at time t = 0, immediately after the pressure p has been applied, the circumferential stress in the inner cylinder is i = (p 3p /4) r /h = pr /4h, and in the outer cylinder the circumferential stress is o = qr /h = 3pr /4h. It is seen that when the pressure is applied (at t = 0), the outer cylinder carries most of the load (q > p q) and o > i). This follows from the fact that the outer cylinder is stiffer than the inner one (Eo > Ei). Because of the lower modulus of elasticity of the inner cylinder, it is easier for the inner cylinder to expand, implying that the outer cylinder (which is less prone to expansion) has to carry most of the load. Due to creep, this stress distribution will be changed as time goes on. (b) How will the circumferential stresses i and o develop with time? The deformation condition ui = uo is valid also at times t > 0. It gives i o o giving + k n = u i = u i Ei Eo

(d,e)

where k is a constant replacing 1 / nn in the material equation (thus, k = 1 / nn), and i > 0 gives sgn i = 1. The dot ( ) indicates differentiation with respect to time t. The contact pressure q between the inner and outer cylinder will now depend on time t, thus q = q(t). The initial condition on q, at time t = 0, was determined above, namely q(0) = 3p /4.
Chapter 1 Page 1:11

Entering i = (p q) r /h and o = qr /h into (e) gives

r n r r + k (p q ) = q (p q ) h Ei h h Eo

(f)

Here p = 0, because p is the constant pressure in the cylinder. Rearranging the expression (f) gives
r n p r n r1 1 + +k q =k (q ) h Ei Eo h h

(g) (h)

or where
C1 =

q + C1 q n = C2 k { r /h } n r { 1/Ei + 1/Eo }/h and C2 = k { p r /h } n r { 1/Ei + 1/Eo }/h

One notices that C2 /C1 = p n ; a relation that will be used below. The homogeneous part of equation (h) reads
q + C1 q n = 0 (i)

This equation may be separated into one part containing q only and one part containing the time t. The solution to (i) becomes, for n > 1,
q = qhom = {( n + 1) ( C1 t + C0 )} 1 / ( n + 1) (j)

where C0 is an integration constant to be determined from a boundary (initial) condition. The particular solution to (h) becomes
C2 1 / n q = qpart = = p C1

(k)

and the complete solution is

q = qhom + qpart = { ( n + 1) ( C1 t + C0 )} 1 / ( n + 1) + p (l)

The boundary (initial) condition q(0) = 3p /4 gives the constant C0. One obtains 3p q (0) = {( n + 1) C0 } 1 / ( n + 1) + p = (m) 4 giving
1 3p 1 p ( n + 1) ( n + 1) = p C0 = n +1 4 n +1 4 (n)

Page 1:12

Chapter 1

Thus,
q
=(n = (n
1 / ( n + 1) 1 p ( n + 1) + 1) C1 t + +p n +1 4 1 / ( n + 1)

p ( n + 1) 1) C1 t + 4
(n

+p

(o)

Finally, one obtains

r r i(t ) = ( p q ) = h h p ( n + 1) 1) C1 t + 4
1 / ( n + 1)

(p)

and
1 / ( n + 1) qr r r p ( n + 1) o(t ) = = (n 1) C1 t + + p h h h 4

(q)

One notices that for t = 0 the initial stresses i = pr /4h and o = 3pr /4h are obtained (as one should). For large values of time t, the term containing the time t will dominate over the second term in the factor enclosed by the curly brackets in (p) and (q). The factor in the curly brackets will then be proportional to t 1 / (n 1), which tends to zero for large values of time t. This implies that after a long time (for large values of time t) the expression giving i in (p), and the corresponding term in (q), will tend to zero, and the stresses tend to i = 0 and o = pr /h, respectively. Thus, the creep in the inner cylinder force the outer cylinder to carry a larger part of the load. After some time the inner cylinder does not carry any load at all (or it carries a very small part of the load). The outer cylinder will then carry the main part of the load. The function of the inner cylinder could then be, for example, to protect the outer cylinder from chemical reactions, or something else. Answer: (a) At time t = 0, the stress in the inner cylinder is i = pr /4h, and in the outer cylinder the stress is o = 3pr /4h. 1 / ( n + 1) (b) r r p ( n + 1) i(t ) = ( p q ) = (n 1) C1 t + h h 4 and
qr r p ( n + 1) o(t ) = = (n 1) C1 t + h h 4
1 / ( n + 1)

r p h

(where i tends to zero and o tends to pr /h when time t becomes large).

Chapter 1

Page 1:13

Fracture, maximum normal stress 1/7. The stress components in a part of a structure have been calculated to xx = 120 MPa, yy = 80 MPa, and xy = 60 MPa (all other stress components are zero). Using the maximum normal stress criterion, investigate material failure. The material is brittle and it has the ultimate strength Ut = 150 MPa in tension and the ultimate strength Uc = 200 MPa in compression. Solution: First, determine the principal stresses in the material (and dont forget the third principal stress). One obtains, in the xy-plane,
xx + yy 1, 2 = 2

xx yy 2 2 + xy = 100 63.2 (MPa) 2

(a)

Thus, the largest principal stress is 1 = 163.2 MPa and the smallest principal stress is 3 = 0 (= zz). Failure is expected, since in tension the ultimate strength of the material is Ut = 150 MPa only. Answer: Yes, failure is expected because the maximum principal stress in the material is 1 = 163 MPa, whereas in tension the strength of the material is Ut = 150 MPa, only.

Page 1:14

Chapter 1

Fracture, Mohrs failure criterion 1/8.

shear stress

The Mohr criterion for brittle fracture is, for the material studied here, formulated as (see figure) + 0.4 = 1.2Y normal where Y is the yield limit of the material. A stress test specimen of this material is subjected to a torque, which is increased until fracture occurs.

Will there be any plastic deformation of the specimen prior to the fracture? Use the Tresca yield criterion. Solution: Pure shear gives Mohrs circle as shown in the figure (centre of the circle at the origine O). Determine the shear stress giving fracture, i.e. determine the radius of the circle. Triangles OAB and OAC give Rfra 1.2Y = 3Y 1.22 + 32 Y which gives Rfra = 1.1142 Y Thus, at fracture one has fra = Rfra = 1.1142 Y. At which shear stress = yield will yielding occur? Yielding will occur when eTresca = Y = 1 3 = yield ( yield) = 2yield. Thus, at yielding one has (according to Tresca) yield = Y /2. It is concluded that yielding will occur before fracture (yielding when = 0.5Y and fracture when = 1.1142Y). This means that the material must be deformation hardening so that shear stress = 1.1142Y can be reached. When that stress is reached the fracture will occur. Answer: Plastic deformation will occur because = fra = 1.114Y at fracture whereas = yield = 0.5Y at yielding.

1.2 Y

C B
Rfra

A
1

3 O

3 Y

Chapter 1

Page 1:15

Fracture, Mohrs failure criterion 1/9. For a brittle material the ultimate strength was determined to be 100 MPa in tension and 400 MPa in compression. Within these limits the material is linearly elastic. Assume that the Mohr criterion for brittle fracture is given by straight lines in the shear stress versus normal stress diagram (as in Problem 1/8). The volume of the test specimens used to determine the ultimate strengths was 104 mm3. This material is used in a circular thin-walled cylinder of length L = 1 m, diameter d = 200 mm, and wall thickness h = 5 mm. The cylinder is loaded with a torque T. With respect to brittle fracture, determine the maximum allowable torque Tmax the cylinder may be subjected to. The size effect according to Weibull should be taken into account. According to Weibull the ultimate strength of a large volume of material is less than the ultimate strength of a small volume of the same material. (One reason for this could be that there is a larger probability of finding a flaw in a larger volume than in a smaller volume. The brittle fracture may start at the flaw.) Assume that the ultimate strength of a material of volume V0 is found to be 0. According to Weibull, the ultimate strength of the same material, but of volume V, would be V0 1 / m = 0 V For the material above, use factor m = 6 in the Weibull relationship. Solution: The ultimate strength in tension, Ut = 100 MPa, and in compression, Uc = 400 MPa, give the fracture limit curve according to the figure below. The shear stress in the circular thin-walled cylinder is T T = = Wv 2 r 2h

(a)

Denote this stress xy. Thus, xy = = T/2r2h. The principal stresses (at plane stress, zz = xz = yz = 0) are given by
xx + yy 1, 2 = 2

where, here, xx = 0, yy = 0, and xy = .

Page 1:16

xx yy 2 2 + xy 2

(b)

Chapter 1

O
Uc

The principal stresses thus are (re-numbered) h 1 = , 2 = 0 and 3 = (c) T Mohrs circle, as given by the principal stresses r 1 and 3, is entered into the diagram. The new circle will have the radius 1 (= OD in the OD = figure). Fracture is expected when the circle E R= 1 reaches the fracture limit curve. The radius of the circle is obtained from triangles OBD and D C B OBC. O O Determine first, however, the distance OB. Ut From triangles OBC and OBE one obtains OB 50 50 = (d) 200 OB + 200 giving OB = 133 MPa.
50 OB 50 = 1 OB (e)

Triangles OBC and OBD give (with OD = 1)

from which 1 = 80 MPa is solved. Thus = 1 = 80 MPa before any reduction of the stress with respect to volume effects has been done. (With no reduction due to volume effects, the allowable torque T would have been T = Wv = 25 kNm.) Reduce the allowable stress with respect to the volume effect. Weibull gives
1/m V0 = 0 V

(f)

Thus,

reduced V0 1 / 6 = V

1/6 104 = 0.3835 = 30.7 MPa giving reduced = (g) 2100 5 1000 The allowable torque T becomes Tallow = reducedWv = 9.6 kNm. Answer: Tmax = 9.6 kNm (and Tmax would have been 25 kNm, approximately, if the size effect had been neglected).

Chapter 1

Page 1:17

Fracture, Mohrs failure criterion 1/10. A circular bar made of a brittle loaded in axial compression. materials are available. Both linear fracture limit curves, see following data e s 2s tan 1/2 1/3 material will be Two different materials have figure, with the

Material A B

Which one of the materials is the better if brittle fracture (in compression) of the bar is considered? Solution:
A
rB rA

B
s

A B

2s

The fracture limit of material A intersects the axis at A = s tan = s / 2. Let the radius of circle A be rA. Similar triangles give rA s /2 (a) = 2 2 rA + s s 1 + (1 / 2) giving rA = 0.809s. This gives UcompA = 2rA = 1.62s.

The fracture limit of material B intersects the axis at A = 2s tan = 2s / 3. Let the radius of circle B be rB. Similar triangles give rB 2s /3 (b) = r B + 2s s 22 + (2 / 3)2 giving rB = 0.925s. This gives UcompB = 2rB = 1.85s. Material B has the largest fracture strength in (uni-axial) compression: UcompB = 2rB = 1.85s. This material should be selected. Answer: Material B should be selected because it has the largest ultimate strength in compression: UcompB = 1.85s (whereas UcompA = 1.62s).

Page 1:18

Chapter 1

Fracture, Mohrs failure criterion 1/11. Two materials, C and D, have linear fracture limit curves. The two materials have the same ultimate strength U when loaded in shear. When loaded in tension, however, material C has a larger ultimate strength than material D, thus UtC > UtD. What can be said about the two materials ultimate strengths in compression? Solution: From the figure it can be seen that the larger the ultimate strength in tension Ut is, the more horizontal will the fracture limit curve be, and the smaller will the circle to the left of the -axis be. Thus, as UtC > UtD one will obtain UcompC < UcompD.

U Ut

It can also be shown that (see below) 1 1 1 = + U Ut Ucomp which gives the above result. Similar triangles give U Ut / 2 1 1 1 = = giving s s Ut / 2 U Ut / 2 s and
Ucomp / 2 U = s s + Ucomp / 2 giving 1 1 1 = + U Ucomp / 2 s

(a)

(b) (c)

Summing (b) and (c) gives (a). Answer: For the ultimate strengths in compression one has (numerically) UcompC < UcompD.

Chapter 1

Page 1:19

Fatigue 1/12. A component of a machine is subjected to a repeated loading sequence. The loading sequence is repeated once a day. The loading is alternating, and one sequence contains the following loading amplitudes (i.e. stress amplitudes the mean value of the stress is zero) a = 180, 160, 140 and 100 MPa The number of loading cycles at each stress level is n = 15, 20, 150, and 500, respectively The Whler curve of the material is, in this stress range, given by the relation a = 50logN + 400 MPa Determine the damage accumulation D due to one loading sequence, and then, determine how many loading sequences (how many days) the component might be used before fatigue failure is expected. Solution: For the different stress levels, the expected fatigue life N is obtained from the Whler curve. One obtains
N = 10
(400 a) / 50

(a)

giving, respectively, N = 25 119, 63 096, 158 489, and 1 000 000 The accumulated damage due to one loading sequence is, by use of the Palmgren-Miner damage accumulation rule, 15 20 150 500 1 D= + + + = 0.00236 = (b) 25 119 63 096 158 489 1 000 000 424 Finally, the number of sequences Ns to expected fatigue failure is Ns = 1 / D 424 Thus, the component can be used in 424 days, approximately. Answer: Damage D due to one sequence is 15 20 150 500 D= + + + 6 = 2.36 10 3 25 119 63 096 158 489 10 Expected number of sequences (days) Ns to fatigue failure is Ns = 1 / D 424.
(c)

Page 1:20

Chapter 1

Elastic instability 1/13.

P x L, EI

(a) First eigenmode: x Pcrit1 (b) Second eigenmode: Pcrit2 x (c) Constrained beam : Pcrit2 x (d)

It is possible to increase the buckling load of a beam by introducing extra supports along the beam, see Section 1.4.9 (the example on instability). It can be shown that by use of one extra support the buckling load of a beam at most can be increase to the second eigenvalue of the original beam without the extra support, i.e. to the second critical load of the unconstrained beam. The extra support should then be placed at the node of the second eigenmode of the unconstrained beam (i.e. at the point of zero deflection of the second eigenmode).

If the extra support is placed at that node, then the beam is constrained to buckle according to the second eigenmode, and thereby the critical load P = Pcrit of the constrained beam (with the extra support) equals the second critical load of the unconstrained beam. Determine where to place one extra support on the beam that is pinned (simply supported) at one end and fixed (rigidly supported) at the other end, as shown in figure (a), to obtain the largest possible critical load of the constrained beam shown in figure (d). In figure (b) the first eigenmode w1(x) of the pinned-fixed beam is sketched, and in figure (c) the second eigenmode w2(x) of the unconstrained beam (a) is sketched. Thus, determine the coordinate x where the mode w2(x) has zero deflection. Solution: The problem will be solved by use of the differential equation for an axially loaded beam. The differential equation reads
EI w IV(x ) + P w (x ) = 0 (a)

where w(x) is the deflection of the beam, EI is the bending stiffness, and P is the force loading the beam in compression. The solution to this equation is w (x ) = C1 + C2 px + C3 sin px + C4 cos px (b)
P / EI . where p =
Chapter 1 Page 1:21

Boundary conditions give the constants C 1 to C 4. One has (BC for boundary condition) BC 1: w (0) = 0 gives C1 + C4 = 0 (c)
BC 2: M (0) = 0 gives EIw (0) = 0 which gives (d)

C3 p 2 sin p 0 C4 p 2 cos p 0 = 0

These two equations give C 4 = 0 and C 1 = 0, which is used in the following boundary conditions. At the beam end x = L one has BC 3: w (L ) = 0 giving C2 pL + C3 sin pL = 0 (e)
BC 4: w (L ) = 0 giving C2 p + C3 p cos pL = 0 (f)

To obtain a solution with at least one of C 2 and C 3 not equal to zero, one must have the determinant of the system of equations (e,f) equal to zero. It gives pL sin pL = 0 or tan pL = pL (g) 1 cos pL
tan pL
pL pL
2
3 2 5 2

The roots to equation (g) are determined numerically. From the figure it is concluded that the first root is close to 3 / 2, the second root very close to 5 / 2, and so on. One obtains p1 L = 4.4934 p2 L = 7.7252 5 /2
p3 L = 7 /2 and so on. (h)

The first root p1 L gives p1 L = P / EI L = 4.4934

which gives 4.49342EI 2.05 2EI P = Pcrit1 = = L2 L2 (i)

The second root p2 L gives p L = P / EI L = 7.7252 2

which gives 7.72522 EI P = Pcrit2 = L2 (j)

The eigenmodes, i.e. the forms of the deflection of the buckling beam, become (equation (f) gives C 2 = C 3 cos pL) for p1 L:
w1(x ) = C3 { p1x cos p1L + sin p1x } (k)

Page 1:22

Chapter 1

for p2 L:

w2(x ) = C3 { p2x cos p2L + sin p2x }

(l)

where C 3 is undetemined (and different in the different modes). The second eigenmode becomes, by use of p2 L (p2 L in radians), x x w2(x ) = C3 7.7252 cos 7.7252 + sin 7.7252 L L

(m)

Now the value of x giving a zero crossing of w2(x) can be determined. One obtains x x (n) w2(x ) = C3 7.7252 cos 7.7252 + sin 7.7252 = 0 L L which gives x = 0.36 L. Thus, if an extra support is placed at x = 0.36 L, the critical load of the pinned-clamped beam increases from Pcrit1 = 2.05 2EI / L 2 = 20.23EI / L 2 (no extra support) to Pcrit2 = 7.72522 EI / L 2 = 59.68EI / L 2 for one optimally placed support. Answer: The support should be placed at x = 0.36 L.

Chapter 1

Page 1:23

Resonance 1/14. A shaft carries two flywheels according to the figure. One of the flywheels is loaded with a torque M(t) = M0 sin t. Determine the maximum torque in the shaft if = 0.75e and = 0.90e, where e is the eigenfrequency of the structure. What happens if = e? The moments of inertia of the two flywheels are 4J and 2J respectively, and the inertia of the shaft can be neglected. Solution: Separate the flywheels from the axle and enter the torque Ma on the axle 2 J M( t ) and on the flywheels. (The inertia of the axle, compared to the flywheels, is supposed to negligible here. Therefore the torque is the same at the two ends of the axle.)

4J

L, GK

2J

M ( x)

4J

Ma

L, GK

Ma

Introduce the rotations 1 and 2 of the two flywheels. The directions of 1 and 2 are shown in the figure. For the flywheel 1 the equation of motion becomes 1 = Ma 4J For the flywheel 2 the equation of motion becomes 2 = M (t ) Ma 2J The torsion of the axle is 2 1. The moment-deformation relationship for the axle is Ma L 2 1 = GK
(a) (b)

(c)

Eliminate 1 and 2 from the equations (a) to (c). That gives a differential equation in the unknown moment Ma. One obtains M M M sint a L + a + a = M (t ) = 0 (d) M GK 4J 2J 2J 2J
Page 1:24 Chapter 1

Assume a particular solution to the differential equation (d) on the form Ma = A sin t

(e)

Enter the assumption (e) into the differential equation (d). It gives an equation determining the unknown amplitude A. One obtains M0 L 3 2 + A= (f) GK 4J 2J from which is solved
A= M0 GK 2JL ( 2 + 3GK / 4JL ) (g)

The (angular) eigenfrequency e can now be determined. The eigenfrequency is the frequency = e giving that the amplitude A in (g) tends to infinity (which is equivalent to that the system determinant in (f) becomes zero). One obtains
e =

3GK 4JL

(h)

The torque in the axle can now be calculated. One obtains, from (e) and (g) 2M0 M0 GK sin t = sint (i) Ma = 2JL ( 2 + 3GK / 4JL ) 3 (1 2/2 e) For = 0.75e one obtains For = 0.90e one obtains
Ma = Ma = 32 M sint = 1.52 M0 sint 21 0 (j) (k)

200 M sint = 3.51 M0 sint 57 0

Finally, if = e, the torque Ma in the axle tends to infinity. It is concluded that an excitation frequency close to a resonance frequency of the structure might be very dangerous. Answer: Normalised with respect to the load amplitude M0, the amplitude Ma of the torque in the shaft is Ma /M0 = Mshaft /M0 = 1.52, 3.51 and , respectively, for = 0.75e, = 0.90e, and = e.

Chapter 1

Page 1:25

Empty page.

Page 1:26

Chapter 1

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 2
Stresses, stress concentration, stresses at crack tip, stress intensity factor, and fracture criteria
Problems with solutions Stress concentration 2/1. A thin-walled circular cylindrical pressure vessel has length L, radius a, and wall thickness h. For a tube connection a small circular hole is opened in the wall (and the tube is mounted). The vessel is then loaded with a pressure p. Determine the maximum stress at the hole due to the pressure if the influence of the tube connection is disregarded. Solution:
tan

long

A B
B long

tan

The stresses in the pressure vessel wall are (tangentially and longitudinally) a a tan = p and long = p (a,b) h 2h Study the stress concentration at the hole. There are four critical points at the hole where high stress may occur: points A and B, see figure. Two points have stress A and two points have stress B. One obtains the stresses
a a a p =p 2h h 2h a a 5 a p = p h 2h 2 h (c) (d)

A = 3 long tan = 3 p and B = 3 tan long = 3 p

The stress around the hole varies between these two values. The largest stress at the hole thus is max = B = 5pa / 2h. Answer: The largest stress at the hole is max = 5pa / 2h.
FMCHAP2P.DOC/2010-01-15/TD

Chapter 2

Page 2:1

2/2. A thin-walled circular cylindrical pressure vessel has a small circular hole in the wall. The hole will be covered with a square plate; the plate being screwed to the wall with four screws, one in each corner. By taking stress concentration into account, where do you want to place the screws in the pressure vessel wall? (This is perhaps a good example to try to understand stress concentration, but it is not a good way to fasten the plate you should never make a hole in a region where you already have stress concentration.) Solution: See answer. Answer: The screws should be mounted as far away from the stress concentration as possible; here it will be at 45o counted from the longitudinal axis of the pressure vessel.

2/3. A large plate has a small elliptical hole in it. The ratio of the axes (major to minor axis) of the elliptical hole is 2 to 1. Determine the stress concentration factor Kt when the plate is loaded in parallel to the (a) minor axis, (b) major axis of the ellipse. Solution: The maximum stress at an elliptical hole is given by a max = 1 + 2 nom (a) b where a and b are the half-axes of the ellipse. (a) If the load is parallel to the minor axis, see Figure (a), one obtains 2 (b) max = 1 + 2 nom = 5nom 1 (b) If the load is parallel to the major axis, see Figure (b), one obtains 1 (c) max = 1 + 2 nom = 2nom 2

max nom nom

(a)

max nom nom

(b)

Answer: Stress concentration factor is Kt = 5 and 2, respectively.

Page 2:2 Chapter 2

2/4. A plate contains an elliptical hole as shown in the figure. The plate is supported at two opposite sides; the support being such that all motion in the y-direction is prevented whereas motion in the x-direction is possible. The plate is loaded with an uni-axial stress along the two other sides.

y b

A
a x

Under which condition (ratio b / a) will fracture start at point A and at point B, respectively? The material is linearly elastic (with Youngs modulus E and Poissons ratio ) up to its brittle fracture at the ultimate strength U. Solution:
yy

y b

A
a x

yy

Due to the prevented contraction in the y direction (strain yy = 0), stress yy will appear in that direction. Let xx = . Hookes law gives 1 yy = 0 = [yy (xx + zz ) ] (a) E Knowing that zz = 0, one obtains yy = xx (b)

Stresses xx (= ) and yy = xx give rise to stresses at the elliptical hole. It is seen that if half-axis b is large (b >> a), then the largest stress at the hole will be at point A. Contrary, if a >> b, then the largest stress will appear at point B. Determine the ratio a / b that gives the same stress at point A as at point B. Using the elementary case for uni-axial stress: a max = 1 + 2 nom and min = nom (c) b one obtains (yy = xx is used)
b A = 1 + 2 xx xx a (d) (e)

and

a B = 1 + 2 xx xx b

Set A = B. It gives b a 1 + 2 xx xx = 1 + 2 xx xx a b
Chapter 2

(f)
Page 2:3

Simplifying (f) gives

1+ b a =0 a b (g) (h)

Let b = a. It gives
2 + (1 ) = 0

Knowing that > 0, one obtains the solution = . Thus, when = , i.e. when b = a, one has A = B. If b > a, then the stress at point A is the largest, whereas b < a gives that the stress at point B is the largest. Answer: Ratio b / a > gives fracture (largest stress) at point A and b / a < gives fracture at point B.

Stresses at crack tip 2/5.

y z r x

Calculate, and show in a graph, the distribution of the maximum shear stress close to a crack tip loaded in Mode I. The material is linearly elastic with Poissons ratio = 0.3.

The stress components close to the crack tip are KI 3 xx = cos 1 sin sin 2 2 2 2 r
yy = 3 cos 1 + sin sin 2 2 2 2 r KI 3 cos sin cos 2 2 2 2 r KI yz = zx = 0 and, at plane deformation, at plane stress zz = (xx + yy ) zz = 0

xy =

Page 2:4

Chapter 2

Solution: Determine the maximum shear stress by use of the Tresca criterion. The principal stresses are then needed. The maximum shear stress is 1 prs max = (prs (a) max min) 2 where superscript prs stands for principal stress. In this problem we have a plane state (either plane stress or plane deformation) in the xy plane. Therefore, the z direction is one principal direction and the stress component zz is one of the principal stresses. In case of plane stress one has the principal stress in the z direction zz = 0
(b)

and in the case of plane deformation (plane strain) one obtains the principal stress KI zz = (xx + yy ) = 2 (c) cos 2 2 r Next, determine the two other principal stresses, called 1 and 2. They are
1 , 2 xx + yy = 2 KI

cos = 2 2 r

xx yy 2 2 + xy 2

3 3 2 cos sin sin + cos2 sin2 cos2 2 2 2 2 2 2

2

= =

KI 2 r KI 2 r

cos 2

2 23 2 3 cos sin sin + cos 2 2 2 2

cos

1 sin 2 2

(d)

The principal stresses 1, 2, and 3 thus become, with = 0.3,

Chapter 2

Page 2:5

1 = 2 =

KI 2 r KI 2 r

cos

1 + sin 2 2 1 sin 2 2

(e) (f) (g) (h)

cos

3 = 0 3 = 0.6

if plane stress KI 2 r cos if plane deformation 2

Plane stress Now the plane stress situation will be investigated. The largest principal stress prs then is prs max = 1 and the smallest principal stress is min = 3 = 0. The maximum shear stress then becomes 1 1 KI (i) max = = cos 1 + sin 2 2 2 2 2 r It is seen that max is a function of . Find the value of that gives the largest max. By solving d max =0 (j) d one finds that max has its largest value when = / 3 = 60 degrees = 60o (and the smallest value of max occurs when = and it is then max = 0). Thus, at plane stress the maximum shear stress will be found in the direction = 60 degrees from the x axis. One has KI (k) max( 60o ) = 0.65 2 r Plane deformation At plane deformation one still has that stress 1 is the largest principal stress. But which one of 2 and 3 is the smallest? To investigate this, set 2 = 3, which gives
if < 47o then 2 > 3 and if > 47o then 2 < 3 (l)

Page 2:6

Chapter 2

Assume that the maximum value of max will be obtained for < 47o. Then max will be given by the principal stresses 1 and 3. One obtains
1 max = (1 3) 2 = 1 KI cos 1 + sin 0.6 cos 2 2 2 2 2 r (m)

Next, find the value of that gives the largest value of max in (m). By solving d max =0 (k) d one finds that max has its largest value when = 75.8 degrees. But this is not in agreement with the assumption made, namely that should be smaller than 47o. Thus, we have to repeat these calculations once again with a new assumption: Assume that the maximum value of max will be obtained for > 47o. Then 2 is the smallest principal stress, and one obtains
1 1 KI max = (1 2) = cos 2 sin 2 2 2 2 2 r (n)

Again, determine the value of that gives the largest value of max. By solving d max / d = 0 one finds that max has its largest value when = / 2 = 90 degrees. This is now in agreement with the assumption that should be larger than 47o. It is then concluded that KI max(90o) = 0.5 (o) 2 r Answer: The maximum shear stress is
at plane stress: max( 60o ) = 0.65 max(90o) = 0.5 KI 2 r KI 2 r

at plane deformation:

Relations (i) and (n) give the graphs asked for (not shown here).

Chapter 2

Page 2:7

2/6. The stress components in front of a crack tip loaded in Mode III are KIII xz = sin 2 2 r
yz = KIII 2 r cos 2

y z

r x

xx = yy = zz = xy = 0

In which direction will the shear stress have its maximum? Solution:
yz

y r

xz

The shear stress on a surface in the xy-plane is KIII 2 = + = sin2 + cos2 xz yz 2 2 2 r

x z

KIII 2 r

= max

(a)

Thus, the shear stress (= max) is the same in all directions (thus independent of the direction ). Answer: The shear stress is the same in all directions.

Page 2:8

Chapter 2

Stress intensity factor 2/7. Study the following crack geometries (a) to (f) and determine the characteristic lengths a of the cracks so that the probability of crack growth is the same for all cracks. All cracks are loaded in Mode I with a remote stress 0. The material is linear elastic. (a) Central crack of length 2a in a large plate. (b) Central crack of length 2a in a quadratical plate with side length 4a. (c) Edge crack of length a in a large plate. (d) Two opposite edge cracks of length a in a rectangular plate with side length (width) 2W = 4a and height h >> W. (e) Half-elliptical surface crack, a deep and 4a (= 2c) long, in a thick plate. (f) Elliptical crack with principal axes a and 2a embedded in a thick plate. Solution: Handbook formulae give stress intensity factors KI (functions fi refer to Appendix 3 in the textbook) 2 (a) KI 2 a h 1 KI KI = 0 a f1 , where f1 = 1.0 gives a = = 0.318 W W 0 0 (b) KI 2 a h KI = 0 a f1 , where f1 = 1.34 gives a = 0.177 W W 0 (c) (d) (e) (f)
a KI = 0 a f5 W a KI = 0 a f4 W a KI = 0 a f7 c a KI = 0 a f8 c where where where where f5 = 1.12 KI 2 gives a = 0.254 0

KI 2 f4 = 1.17 gives a = 0.233 0 KI 2 f7 = 0.896 gives a = 0.397 0 KI 2 f8 = 0.826 gives a = 0.467 0

Answer: Ratio a / (KI / 0)2 is (a) 0.318; (b) 0.177; (c) 0.254; (d) 0.233; (e) 0.397; (f) 0.467.

Chapter 2

Page 2:9

Fracture modes 2/8. A thin-walled circular cylinder contains a crack. R The plane of the crack is inclined an angle (= N T ) with respect to the longitudinal axis of the cylinder, see figure. The cylinder is loaded by a torque T and an axial tensile force N.
t

Determine a relationship between T, N and so that the crack is loaded in Mode I only, and in Mode II only. Assume that R >> t so that the curvature of the cylinder may be neglected when the stress intensity factor is determined. Solution:
n

area = 1
xy

sin cos
xy

xx

yy

The axial tensile force N gives the normal stress xx in the longitudinal direction, and the torque T gives shear stress xy, where N T (a,b) xx = and xy = 2Rt 2R 2t The stress yy is zero (yy = 0).

Equilibrium equations in the directions of n and n give (in the equations below the area is written after the multiplication sign , i.e. the area equals 1 for the largest area and sin and cos , respectively, for the two smaller areas) n 1 = xx sin sin + 0 xy sin cos xy cos sin
2T N cos sin + = xx sin2 2xy cos sin = sin 2Rt 2R 2t (c)

and

n 1 = x sin cos 0 + xy sin sin xy cos cos = N T cos 2 sin 2 + 4Rt 2R 2t (d)

The loading of the crack is in Mode I if n 0 and n = 0, which gives T N T R n = 0 = cos 2 giving sin 2 + = tan 2 4Rt N 2 2R 2t Loading of the crack is in Mode II if n = 0 and n 0, which gives, either

(e)

Page 2:10

Chapter 2

n = 0 = sin giving = 0

(f) T R = tan N 2

or
2T N cos sin + n = 0 = sin 2Rt 2R 2t giving (g)

Answer: Load ratio T / N = ( R / 2) tan 2 gives Mode I loading of the crack, and ratio T / N = ( R / 2) tan gives Mode II loading.

Linear elastic fracture mechanics (LEFM) 2/9. In a thin-walled pressure vessel there is a risk that a 300 mm (= 2a) long longitudinal through2R 2a thickness crack will develop almost immediately as the pressure vessel is loaded (due to a bad t welding, for example). In an approximate l estimation of the stress intensity factor the curvature of the pressure vessel wall was disregarded. The stress intensity factor KI was then determined for a crack in a plane, infinitely large plate, and it was found that the stress intensity factor so determined was only half of the fracture toughness KIc of the material. What would a more accurate analysis give? Could the pressure vessel be used without danger? The vessel is manufactured of steel, and it is loaded with a pressure p. The conditions for linear elastic fracture mechanics are assumed to be fulfilled. Numerical data: pressure vessel radius is R = 0.4 m, overall length of pressure vessel is l = 5 m, crack length is 2a, where a = 0.15 m, and the wall thickness is t = 0.015 m. Solution: The stresses in the wall are
rr = 0 , = pR , t zz = pR , and all ij = 0 2t (i j ) (a)

The requirements for linear elastic fracture mechanics (LEFM) to be valid are assumed to be fulfilled.

Chapter 2

Page 2:11

Case A. Study a crack in a large flat plate. The stress intensity factor then is pR KI = 0 a f1(0, 1) = 0 a 1 = a (b) t
But KI = KIc 2 gives KIc = 2 pR a t (c)

The fracture toughness KIc of the material has now been determined. Case B. Study a crack in a cylindrical wall. The stress intensity factor will now be obtained from a handbook (Case 14 in Appendix 3 of the textbook). The following conditions must be fulfilled: R >> t, a >> t and L >> R. They are all fulfilled. Determine . One has
= a Rt 0.4 0.015 = 0.15 = 1.936 KI = 2.343 pR a t (d) (e)

This gives f14(1.936) = 2.343. Thus

which is larger than the fracture toughness KIc and fracture will be expected! Answer: Exact analysis (taking curvature into account) gives KI 1.17KIc, implying that failure will occur.

2/10.
M0

2a
M0 t

A through-thickness crack of length 2a has been found in a large plate. The plate is subjected to a bending moment M0 (Nm/m) per unit length. Determine at which moment M0max crack growth will occur.

Numerical data: a = 0.02 m (crack length 2a), plate thickness t = 0.03 m, yield strength Y = 1300 MPa, and fracture toughness KIc = 110 MN/m3/2.

Page 2:12

Chapter 2

Solution: Case 13 in the Appendix 3 of the textbook gives 6M 0 a KI = 0 a f13 where 0 = 2 and M0 in Nm/m W t Ratio a / W = 0 gives f13 = 1. Thus
KI = 6M 0 t2 a

(a) (b)

Crack propagation is expected when KI = KIc if linear elastic fracture mechanics (LEFM) can be used. One has
KIc 2 110 2 = 0.0179 m 2.5 = 2.5 1300 Y (c)

which is smaller than a, t and W a. Thus LEFM can be used. The fracture criterion KI = KIc gives from which M0 is solved. One obtains
M0 = KIc t 2 6 a = 110 0.032 6 0.02 = 0.0658 MNm m (e) 6M0 t2 a = KIc (d)

This value of the moment M0 gives the stress in the plate. One obtains 6M0 = 2 = 439 MPa = 0.34Y t

(f)

Answer: Crack growth is expected when bending moment of the plate is M0max = 66 kNm/m, giving maximum remote stress = 0.34 Y in the plate. 2/11. A cantilever beam contains a through-thickness crack. The beam is loaded by a system of forces 11 P d (11P and 9P) according to the figure. Determine d at which value of P failure may be expected. 9P Numerical data: a = 0.02 m (crack length 2a), beam (plate) thickness t = 0.03 m, plate width b = 0.08 m, d = 0.06 m, l = 3.2 m, yield strength Y = 600 MPa, and fracture toughness KIc = 50 MN/m3/2.

b 2a l

Chapter 2

Page 2:13

Solution: The cantilever beam (or plate) will be loaded in tension and in bending. The axial force becomes N = 20P loading the plate in tension, and the bending moment becomes M = 2Pd. The stress intensity factor KI is obtained by superposition of the two loading cases. One obtains (Cases 1 and 13 in Appendix 3 of the textbook) a h 6M a N KI = KIN + KIM = a f1 , + 2 a f13 W W b t W bt
= = 0.02 1.6 12Pd 0.02 a 20P f1 , + f13 bt

0.04 0.04

0.04

P d a 20 1.186 + 12 1.162 bt t P a = 51.6 P a bt bt (a)

= (23.72 + 27.89)

If the conditions for linear elastic fracture mechanics, LEFM, at plane strain are fulfilled, then failure will occur when (b) KI = KIc Can LEFM be used?
KIc 2 50 2 = 0.0174 m 2.5 = 2.5 600 Y (c)

This number (0.0174 m) is less than the crack (half-)length a, plate thickness t, and the distance b / 2 a from the crack tip to the edge of the plate, so LEFM can used. Thus,
KI = 51.6 P a = K = 50 MN/m3 / 2 Ic bt (d)

giving
P = Pcrit = KIc bt 51.6 a = 50 0.08 0.03 51.6 0.02 = 0.00928 MN = 9300 N (e)

Failure is expected at P = Pcrit = 9.3 kN. Answer: Failure is expected at critical load Pcrit = 9.3 kN
Page 2:14 Chapter 2

2/12.
M0 P0 t a W P0 M0

An edge crack has been discovered in a long beam, see figure. The fracture toughness KIc and the yield strength Y of the material depend on temperature as given below.

(a) Over which temperature range may linear elastic fracture mechanics be used? (b) For the upper limit of this temperature range, determine at which loadings P0 and M0 fracture may be expected. Numerical data: crack length a = 0.1 m, beam height W = 0.5 m, beam width t = 0.2 m.
Y

MPa 700 600 500 400 -200 -100 0 Temperature T oC

Y

K Ic

K Ic MN/m 3/2 160 140 120 100 80 60 40

100

Figure. Fracture toughness KIc and yield strength Y as function of temperature T Solution: (a) First, investigate if linear elastic fracture mechanics, LEFM, can be used. One has
KIc 2 2.5 minimum of (a , t , W a ) = a Y which gives KIc Y

Thus, numerically, one should have KIc < 0.2Y. This condition is fulfilled if temperature T is T 3oC, see diagram (at T = 3oC, approximately, one obtains KIc = 100 MN/m3/2 and Y = 500 MPa).
Chapter 2 Page 2:15

a = 0.2 ( m) 2.5

(a)

(b) Superposition of the two loading cases tension and bending gives (Case 5 and 6 in the Appendix 3 of the textbook) P0 a 6M 0 a KI = KIP + KIM = a f5 + a f 6 W t W2 W bt
= = a 0.1 6M0 0.1 P0 f5 + f tW 0.5 W 6 0.5 tW (b)

a (1.366P + 12.902M ) 0 0

Fracture will occur when

KI = KIc (c)

At temperature T = 3oC the fracture toughness KIc is KIc = 100 MN/m3/2. Thus, at this temperature fracture will occur when
KI = a (1.37P + 12.90M ) = K 0 0 Ic tW KIc tW a = 100 0.2 0.5 0.1 = 17.84 MN (d) (e)

giving

(1.37P0 + 12.90M0) =

with P0 in MN and M0 in MNm. Thus, at fracture one has

P0c + 9.44M0c = 13.1 MN (f)

where P0 should be entered in MN and M0 in MNm. Answer: Temperature T should be T 3oC. At temperature T = 3oC the yield strength is Y = 500 MPa and the fracture toughness is KIc = 100 MN/m3/2 giving the maximum load P0 + 9.44M0 = 13.1 MN, where P0 should be entered in MN and M0 in MNm.

Page 2:16

Chapter 2

2/13. A thin-walled circular cylinder contains a crack o T forming an angle 30 to the longitudinal axis of the cylinder, see figure. Determine at which torque T crack propagation will be expected. In this case, the curvature of the wall may be neglected when the stress intensity factor is estimated (i.e., the radius R is very large compared to crack length a and wall thickness t).
4 2 KIII +1

30 o

R t

The equivalent stress intensity factor Ke is

2 Ke2 = KI2 + KII +

Numerical data: radius R = 0.2 m, crack (half-)length a and wall thickness t are a = t = 0.005 m, yield limit Y = 1200 MPa, and fracture toughness KIc = 50 MN/m3/2. (For a small crack in a large plate the stress intensity factor in Mode II is KII = x y a g , where (here) g = 1.) Solution: Can linear elastic fracture mechanics, LEFM, be used? One has
KIc 2 50 2 (a) = 0.00434 m 2.5 = 2.5 1200 Y Here a = t = 0.005 m > 0.00434 m (and W a is larger), so LEFM can be used.
xy

n
xy

60 o 30
o

Consider an element surrounding the crack, see figure. The stress on the element is T T (b) xy = = Wv 2R 2t where Wv is the section modulus in torsion of the cylinder. The stress xy gives a normal stress n across the crack, loading the crack in Mode I, and a shear stress n, loading the crack in Mode II.

Chapter 2

Page 2:17

The stresses n and n become, with = 60o,

n = xx cos 2 + yy sin2 + 2xy cos sin =

3
2

xy

(c) (d)

1 n = (xx yy ) sin cos + xy (cos2 sin2 ) = xy 2

In Mode I the stress intensity factor KI becomes (for a small crack in a large plate the factor f1 becomes f1 = 1)
KI = n a f1 =

3
2

xy

(e)

In Mode II the stress intensity factor KII becomes (for a small crack in a large plate the factor g becomes g = 1) 1 KII = n (f) a g = 2 xy a The equivalent stress intensity factor Ke becomes 3 1 2 2 = 2 Ke2 = KI2 + KII xy a + = xy a 4 4 Fracture is expected when Ke = KIc (if LEFM can be used). This gives T Ke = xy a = a = KIc 2R 2t Thus,
T= KIc 2R 2t a = 50 2 0.22 0.005 0.005 = 0.501 MNm

(g)

(g) (h)

At torque T = 500 kNm crack propagation is expected. (The LEFM conditions were checked above.) Answer: Crack propagation is expected when torque T = 500 kNm.

Page 2:18

Chapter 2

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 3
Small plastic zone at crack tip
Problems with solutions The Irwin correction 3/1. At a circular hole (radius r = 10 mm) in a large plate two cracks (length a = 5 mm) have been discovered, see figure. Determine the maximum remote stress (the critical stress) the plate may be loaded with. Data: plate thickness t = 20 mm, fracture toughness KIc = 40 MN/m3/2, and yield limit Y = 500 MPa.

r a a

Solution: Can linear elastic fracture mechanics be used?

KIc 2 40 2 2.5 = 2.5 (a) = 0.016 m 500 Y The plate thickness t is large enough to ensure plane strain, but the crack length a is too short. Therefore, use the Irwin correction of the crack length. For plane strain (t is large enough), the correction at critical stress, i.e. when KI = KIc, is
2 1 KIc 1 40 2 r1 = = = 0.00034 m 6 Y 6 500

(b)

FFFCH3P.DOC/2010-01-15/TD

Chapter 3

Page 3:1

Correction of LEFM now gives

a + r1 KI = ( a + r ) f 0.00534 1.78 = 1 3 r + a + r1 = KIc = 40 106 N/m3 / 2 (c)

from which = crit = 173 MPa is solved. Answer: The critical remote stress is, approximately, = crit = 173 MPa.

3/2. A beam with rectangular cross section, t = 20 mm and W = 40 mm, is loaded with a bending moment M as shown in the figure. A 12 mm deep crack (a = 12 mm) has appeared in the beam (see figure). In order to increase the load-carrying capacity of the cracked beam, the material around the crack is removed. It is decided that 9 mm of the beam height should be removed by grinding. After the grinding the beam height thus is 31 mm. The crack, however, is still there, and its depth is now 3 mm. Material properties: KIc = 40 MN/m3/2 and Y = 900 MPa.
removed

9 mm

(a) How much has the ultimate bending moment been increased by the grinding? (Thus, calculate the ultimate load M of the beam before grinding and after the grinding.) (b) The purpose of the grinding was perhaps to remove all the crack. What bending moment can be put on the beam if 12 mm were removed (i.e. even the crack tip has been removed)? In this case the yield limit of the material limits the loading.

Page 3:2

Chapter 3

Solution: Can LEFM (linear elastic fracture mechanics) be used?

KIc 2 40 2 2.5 = 2.5 (a) = 4.94 103 m = 4.94 mm 900 s This length, 4.94 mm, is less than t (= 20 mm), a (= 12 mm), and W a (= 28 mm), respectively. Thus, it is OK to use LEFM for the first crack length (12 mm) but not for the second (length 3 mm).

Determine the ultimate load before grinding, i.e. the ultimate bending moment for the beam with cross section 20 mm by 40 mm and crack depth 12 mm: Case 6 in Appendix 3 of the textbook gives
KI = a 0.012 6M 6M 0.012 f6 = a f6 2 2 W 0.020 0.040 0.040 tW (b)

= KIc = 40 106 N/m3 / 2

Diagram gives f6 = 1.115, and one obtains the ultimate load M = M1 = 985 Nm. After the grinding the cross section becomes 20 mm by 31 mm, and crack depth is 3 mm. LEFM can not be used any longer. Use the Irwin correction of the crack length. One obtains for plane strain, when the stress intensity factor reaches KIc,
2 1 KIc r1 = = 0.000105 m 6 Y

(c)

Case 6 gives aeff 6M 0.003105 6M = a f 0.003105 f KIeff = eff 6 6 0.031 W 0.020 0.0312 tW 2
= KIc = 40 106 N/m3 / 2 (d)

Diagram gives f6 = 1.057, giving M = M2 = 1227 Nm. (b) If all the material surrounding the crack had been removed, i.e. if 12 mm of the beam height were taken away (this was perhaps the original purpose of the grinding operation), then fracture mechanics theory needs not be used at all. In that case, determine the bending moment the beam can be loaded with, if plastic deformation of the beam should be avoided. One obtains

Chapter 3

Page 3:3

max =

M W /2 M 0.028 / 2 = Y = 900 MPa = I 0.020 0.0283 / 12

(e)

which gives M = 2352 Nm. (Conclusion: make sure that also the crack tip is removed, if you try to improve the load-carrying capacity of a structure by removing a crack.) Answer: (a) The maximum bending moment M increases from 985 Nm to 1227 Nm, thus by 25 per cent. (b) The moment can be increased to M = 2352 Nm if the full crack (with crack tip included) is removed.

The Dugdale model 3/3. A large steel plate of an elastic, ideally plastic material (Y = 500 MPa) contains a throughthickness crack of total length 50 mm (= 2a). At a tension test of the plate the crack was found to start growing at the remote stress = 300 MPa. (a) At which remote stress would a 150 mm long crack start to grow? Use the Dugdale model of crack tip opening displacement (a) as criterion for crack growth initiation (CTOD criterion). One has 8 a Y 1 (a ) = ln E cos ( / 2Y)

(a)

y x

2a

(b) For the two cases 2a = 50 mm and 2a = 150 mm, calculate the size (length) of the plastic zone at the crack tip when the stress is such that the crack starts to propagate. Solution: The first measurement, at crack length 2a = 50 mm, gives the critical value of the crack tip opening displacement (CTOD) (a)crit. One obtains

Page 3:4

Chapter 3

(a )crit = =

8 acrit Y 1 ln E cos ( crit / 2Y)

8 0.025 Y Y 1 = 0.03382957 ln E E cos ( 300 / 2 500)

(a)

(a) At which stress would a 150 mm long crack start to grow? The CTOD criterion gives, with (a)crit just calculated in (a), Y 8 0.075 Y 1 ln (a )crit = 0.03382957 = E E cos ( / 2 500) Solving for gives = 184 MPa. (b) The length of the plastic zone is obtained from 1 1 =a cos ( / 2 Y) For 2a = 50 mm, one obtains, with remote stress = 300 MPa,
1 = 0.025 1 = 0.0175325 m = 17.5 mm cos ( 300 / 2 500)

(b)

(c)

(d)

For 2a = 150 mm, one obtains, with remote stress = 184 MPa, 1 = 0.075 1 = 0.0145492 m = 14.5 mm cos ( 184 / 2 500)

(e)

Answer: (a) At crack length 150 mm the critical stress is 184 MPa, (b) at crack length 50 mm the length of the plastic zone is 17.5 mm (for critical remote stress 300 MPa), and at crack length 150 mm the length of the plastic zone is 14.5 mm (for critical stress = 184 MPa).

Chapter 3

Page 3:5

3/4. During a tension test of a large plate with a central crack the crack opening (0) was recorded as a function of the load. The recorded curve deviated from a straight line. The deviation from linearity may be explained partly by the appearance of a plastic zone at the crack tip and partly by crack growth during the loading.

(0)

y x

2a

(a) How large deviation from linearity will be expected due to the plastic zone at the crack tip? The remote stress is Y / 2. (b) The deviation from linearity of the recorded curve appeared to be 30 per cent. Calculate the crack propagation during the loading (assume that symmetry is maintained, i.e. the two crack tips move the same distance). The material is linearly elastic, ideally plastic with yield strength Y. Plane stress is at hand so that the Dugdale model may be used. The crack opening displacement (COD) is (0) 8Y 1 + sin( / 2Y) = ln a E cos( / 2Y) and if << Y one obtains
4 KI (0) = a E a
2 2 KI 1 + + .. 24 a Y

Solution: (a) Using Dugdales model, the crack opening displacement (0) becomes (0) 8Y 1 + sin( / 2 Y) = ln a E cos( / 2 Y)
= 8 2 1 + sin( / 4 ) ln = 4.4888 E cos( / 4) E (a)

If << Y (i.e. if Y is very large) one has

Page 3:6

Chapter 3

2 4 KI (0) 2 KI 1 + = + .. a 24 Y E a a

(b)

Using KI = a and << Y in (b), one obtains the linear elastic solution. (If Y tends to infinity, no yielding will occur, and the solution obtained is for the fully elastic case.) This gives
a (0) 4 = (c) { 1 + 0} = 4 a E E a The difference between the displacements in (a) and (c) comes from the plastic deformation at the crack tip. The deviation from linearity is 4.4888 4 = 0.1222 = 12.2 per cent (d) 4

(b) At the experiment, the deviation from linearity was 30 per cent. The plastic deformation at the crack tips gives 12.2 per cent deviation from linearity only, so the remaining deviation up to 30 per cent is explained by crack growth; i.e., the crack has propagated during the loading of the structure. Calculate the final crack length afinal. Using (0) = 1.3 elastic, where elastic was calculated in (c), the Dugdale model gives a 8 2 afinal 1 + sin( / 4 ) = ln (e) (0) = 1.3 4 E E cos( / 4 ) from which is solved
afinal = 1.3 4 a = 1.158 a 4.4888 (f)

Answer: (a) Deviation from linearity due to plastic zones is 12.2 per cent, and (b), crack has grown to final length afinal = 1.158a during the loading, where a is the original crack length in the unloaded structure.

Chapter 3

Page 3:7

3/5.

(0)

2a 2W

During a tension test of a large plate with a central crack the crack opening (0) was recorded as a function of the load P. The recorded P curve deviated from a straight line as given in the figure. Symmetric stable crack growth was x 2 h observed during the loading of the plate. Calculate the crack growth during the loading. The material is linearly elastic, ideally plastic with yield strength Y. The plate is thin so that the stress state is plane (PS). Thus, the Dugdale model may be used. Numerical data: crack length a0 = 0.01 m, plate thickness t = 0.002 m, plate size h = W = 0.2 m, yield limit Y = 600 MPa, and modulus of elasticity E = 200 GPa.

P P MN

0.28

(0 ) 0.18 mm

Solution: At fracture the remote tensile stress is

c =

Pc 2Wt

(a)

The crack opening displacement (COD) at fracture is 8 Y ac 1 + sin ( /2Y) ln c(0) = E cos ( /2Y) which gives
1 + sin (Pc /4Wt Y) 1 ac = E c(0) 8 Y ln cos (Pc /4Wt Y)

(b)

(c)

According to the diagram, fracture occurred when Pc = 0.28 MN and c(0) = 0.1810 3 m, which gives ac = 0.0218 m, giving 2ac = 0.0436 m During the loading of the plate, the crack has grown the amount 2a = 2ac 2a0 = 0.0436 0.0200 m = 0.0236 m = 23.6 mm. (e) (d)

Answer: The crack growth is 2a = 23.6 mm (thus, the crack has propagated 11.8 mm at each crack tip).
Page 3:8 Chapter 3

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction Studentlitteratur, Lund 2002, ISBN 91-44-02096-1

Chapter 4
Energy considerations
Problems with solutions Surface energy, stresses at crack tip 4/1. The surface energy ws and the cohesive strength c of a material may (approximately) be x determined from the following simple model of i the inter-atomic forces: Assume that the force i per unit area between i atomic planes may be represented by the c function x 5x0 x0 x x0 i = c sin 4 x0 where x0 is the distance between the planes of the atoms in the unloaded state and x is the atomic separation when the material is loaded. The factor c is a material parameter: the bond strength. Also, assume that the fracture limit is obtained when the strain is 25 per cent, i.e. when x = 5x0 /4.
i

(a) Determine the parameters , c and ws in terms of E and x0, where E is the modulus of elasticity of the material. (b) Compare the fracture criterion by Griffith (released energy equals surface energy) with a stress criterion for fracture of a material with a crack. Use a small through-thickness crack in a large plate and compare the remote stresses giving crack growth in the two cases.

FFFCH4P.DOC/2010-01-15/TD

Chapter 4

Page 4:1

For the stress criterion, assume that the mean stress over n planes of atoms in front of the crack tip is a measure of the loading of the material. Determine how far away in front of the crack tip (measured in number of atomic planes) the mean stress must be equal to the cohesive strength of the material in order to make the stress criterion equivalent to the energy criterion by Griffith. (The stress state may be considered as plane.) Solution: (a) At small displacements (i.e., when x is close to x0), Hooke s law gives x x0 i = E = E (a) x0 which gives
d i E = d x x0 i = c sin x x0 x0 (c) (b)

From the equation given in the problem formulation one obtains

d i x x0 cos = c dx x0 x0 E

Equations (c) and (b) give, for x = x0, di(x = x0) E cos 0 = = c dx x0 x0

giving

c =

(d)

From the problem formulation, see figure above, one has, when x = 5x0 /4, d i =0 dx which gives Thus,
d i(x = 5x0 /4) 5 x0 /4 x0 cos =0 = c dx x0 x0 1 cos = 0 4 1 which gives = and = 2 4 2

(e) (f) (g)

Enter = 2 into (d). It gives the cohesive stress (cohesive strength) E c = 2

(h)

Page 4:2

Chapter 4

It is noted that this stress is very high: in the order of E / 10. The ultimate strength U of a very high-strength steel is in the order of E / 100, which means that in practice, fracture will occur due to other mechanisms than separation of the atomic planes. The surface energy may now be determined from the area below the curve i(x). One obtains
2ws = x0
3x0 /2

i(x ) dx = x0

3x0 /2

x x0 E dx sin 2 2 x0
3x0 /2

x x0 E x0 = cos 2 2 2 x0 x

=
0

E x0 2 2

(i)

where the factor 2 in front of ws is there because two new surfaces are created when the crack appears. Thus E x0 [Nm / m2 ] (j) ws = 2 4 (b) The Griffith criterion According to Griffith, the energy that is released when a crack propagates should be equal to the energy needed to create the new crack surfaces. By use of the stress intensity factor KI, this crack propagation criterion can be written, see equation (4.34) in the textbook, +1 2 (k) K t a = 2ws t a 8 I where = (3 ) / (1 + ) at plane stress, and is the shear modulus of the material: = E / 2(1 + ). The crack growth is a (the crack length a has increased by a) and the plate, in which the (through-thickness) crack is situated, has thickness t. Expression (k) is thus given for the new crack area ta created. Solving (k) for KI gives
KI =

2ws 8 = +1

2ws 8E
3

2(1 + ) 1 + + 1

= 2w sE

(l) (m)

Thus, fracture will occur when

KI = Kc, where Kc = 2w sE

Chapter 4

Page 4:3

Now, determine the remote stress G (superscript G for Griffith) leading to failure. For a small through-thickness crack in a large plate, the stress intensity factor KI is
KI = G a KI = G a = Kc = 2w sE (n) (o)

Crack propagation will occur (for plane stress conditions) when KI = Kc, giving Thus, according to the Griffith criterion the crack will propagate when the remote stress is
G =

2ws E = a
r

2E E x0 E = a 42

x0 2a

(p)

The stress criterion

y
yy =

KI

We now turn to the stress criterion. The stress yy in front of the crack tip is
yy = KI 2 r = a 2 r =

a 2r

(q)

where is the remote stress (far away from the crack). The coordinate r is used here in order to avoid confusion with the distance x, or x0, between the atomic planes. The mean value yy of yy = yy(r) over n atomic planes is
yy 1 = n x0 0 =
n x0

1 yy (r ) dr = n x0 0

n x0

a dr 2r (r)

n x0 a 1 r ] [ 2 0 = n x0 2

2a n x0

At fracture the mean value yy should reach the cohesive strength c of the material. This gives
yy =

2a E = c = n x0 2

(s)

Page 4:4

Chapter 4

and

= S =

E 2

where superscript S stands for stress criterion.

n x0 2a

(t)

We have now determined the stress G giving failure according to the Griffith criterion and the stress S giving failure according to the stress criterion. In the case that these two criteria give failure at the same remote stress, one has G = S . This gives
G = E

which gives n. Thus n = 4/ = 1.27.

x0 E = 2a 2

n x0 = S 2a

(u)

In conclusion, if the mean stress yy of yy(r) reaches the cohesive strength c over n = 4/ atomic planes, then the stress criterion (as formulated here) and the Griffith criterion will predict crack propagation at the same remote stress . The material properties should then be as given in the equation in the problem: i = c sin (x x0) / x0, with parameter values as calculated in problem (a). Answer: (a) = 2, c = E /2, ws = E x0 /42, (b) n = 4 / = 1.27.

Tensor notations, strain energy 4/2. Determine the elastic strain energy density in front of a crack tip loaded in mode III. In which x direction has the strain energy its maximum? z The elastic strain energy density u is, for linearly elastic material, ij ij u= 2 The stress components in front of a crack tip loaded in mode III are KIII KIII yz = xz = sin ; cos 2 2 2 r 2 r
and xx = yy = zz = xy = 0
Chapter 4 Page 4:5

Solution: Use

u=

ij ij 2 ij = 1 2 ij

(a) (b)

Calculate the strains. One has, for i j,

where is the shear modulus. Enter the stresses given in the problem, and the strains calculated in (b) into the expression (a). It gives
2 zx yz zy 1 KIII 1 xz 2 sin2 + 2 cos2 u = xz + zx + yz + zy = 2 2 2 2 2 2 2 4 2 r 2 1 KIII 1 2 = = KIII 2 2 r 4 r

(c)

Answer: The elastic strain energy u is u = KIII2/4r (thus independent on ), is the shear modulus.

Energy balance 4/3.

P

a
Y

Two rectangular bars (same cross section b by h) of linearly elastic material (modulus of h elasticity E) are glued together along a portion h of its length, see figure (a). The glue has - b (a) characteristics according to figure (b) (rigid, ideally plastic up to deformation U and yield limit Y). Determine the critical load P = Pcrit. One has a >> h and a >> b.
(b)

Solution: Two solutions will be given. (a) Load control gives the potential energy = U P
where which gives
Page 4:6

(a) (b) (c)

ij ij P U = dV = 2 Vol 2 = P / 2
Chapter 4

Assume that the two rectangular bars may be considered to be cantilever beams of length a. The displacement /2 of the cantilever beam end due to a force P then is P a3 2 P a3 = giving = (d) 2 3 EI 3 EI The potential energy of the structure becomes
P2 a3 P = = 2 3 EI (e)

The energy release due to a change of the crack area is P 2 (b a )3 P 2 3(b a )2 P 2 a 2 d d d = = = = d A d (b a ) d (b a ) 3EI b 3 3 EI EI b b3

(f)

The work required to create the crack surface, i.e. the work required to overcome the glue strength, is Ws = A Y U ( = A 2ws) (g)
which gives d Ws = Y U dA d d Ws = , dA dA EI b Y U a
2

(h) gives P2 a2 = Y U EI b = bh a E h Y U 2 3 (i) (j)

Energy balance, namely that gives

P = Pcrit =

= 12 a
2

E b h 3 b Y U

(b) Alternative solution

From P2 a3 = 3 EI P2 3 a2 = a = a a 3 EI and Ws = b a Y U Ws a = b Y U a a gives = bh a P2 a2 = b Y U EI (k,l)

one obtaines, for a crack growth a,

and Ws = (m,n) (o) (p)

Energy balance, namely

= Ws ,

Thus

P = Pcrit =

Answer: Critical load is

a2

EI b Y U

E h Y U 2 3 E Y Uh

Pcrit =

bh 2a 3

Chapter 4

Page 4:7

4/4.
P

To determine the surface energy ws of a brittle, linearly elastic material a test specimen, see figure, was manufactured from the material to be investigated. When loading the test specimen 2 h crack propagation was obtained when the force P was P = Pmax. Determine the surface energy of b the material. The modulus of elasticity is E, the cross section is b by 2h, and the crack length is a where a >> h and a >> b.

Solution: Two solutions will be given. (a) Load control gives the potential energy P P = U P = P = 2 2

(a)

/2
P a

Assume that the rectangular bars may be approximated to two cantilever beams of length a. The displacement /2 of the beam end due to a force P is
P a3 2 P a3 = giving = 2 3 EI 3 EI P 2 (ba )3 P = = 2 3 EI b 3 (b)

The potential energy of the structure is

(c)

The energy release due to a change of the crack area A = b a is

d P2 a2 d P 2 3(ba )2 = = = d A d (ba ) 3 EI b 3 EI b (d)

The work required to create the new surfaces is

Ws = 2ws A which gives d d Ws = dA dA d Ws = 2 ws dA gives P2 a2 = 2ws EI b (e) (f) (g)

Energy balance which gives

(G =)

2 2 P2 a2 P 2 a 2 12 6 Pmax a = = ws = 2 b EI 2 b E bh 3 E b 2 h 3

Page 4:8

Chapter 4

Thus, the surface energy ws is ws = 6Pmax2 a 2 / E b 2 h 3. (b) Alternative solution

From P2 a3 = 3 EI and W s = 2 b a ws (h,i)

is obtained, for a crack growth a,

Ws P2 3 a2 = a = a and Ws = a = 2 b ws a a 3 EI a Energy balance, = Ws, gives P2 a2 = 2 b ws EI Thus ws = P2 a2 2 b EI (j,k)

(l)

Answer: The surface energy ws is ws = 6Pmax2 a 2 / E b 2 h 3. 4/5.

M

A thin rectangular bar (cross section b by h) of b linearly elastic material (Youngs modulus E) is h glued to a rigid foundation, see figure (a). The (a) glue has a linear - characteristics according to figure (b) (rupture at stress U when deformation is U). The bar is loaded with a bending moment M. Determine the critical load Mcrit. One has a >> b > h.
(b)

Solution: Two solutions will be given. (a) Load control gives the potential energy = U P, which here gives M M =U M = M = 2 2

(a)

Assume that the free part of the rectangular bar may be approximated to a cantilever beam of length a. The rotation of the beam end due to the bending moment M is M a M a 12 = (b) = EI E bh 3 The potential energy of the structure becomes (crack area A is A = ba)
M M 2 (ba ) 6 M2 a 6 = = = 2 E bh 3 E b 2h 3
Chapter 4

(where A = ba )

(c)
Page 4:9

The energy release due to a change of the crack area is

d M2 6 = dA E b 2h 3 (d)

The work required to create the new surface is

1 Ws = A U U 2 which gives d Ws 1 = dA 2 U U (e)

Energy balance,
d d Ws M2 6 1 = U U = gives dA dA E b 2h 3 2 (f)

which gives
2 M 2 = Mcrit =

U U bh E b 2h 3 and Mcrit = 12 2

U U E h 3

(g)

(b) Alternative solution

From M2 6 a = E b h3 6 M2 = a a = a E b h3 and 1 Ws = ba U U 2 Ws 1 a = b U U a a 2 (h,i)

is obtained
and Ws = (j,k)

Energy balance, = Ws, gives

6 M2 1 = b U U E b h3 2 (l)

Thus
M =M
2 2 crit

E U U b 2h 3 N N 2 3 2 dimension: 2 2 m m m = (Nm) = 12 m m

(m)

as obtained in (g). Answer: Critical load Mcrit is

Mcrit = bh 2 3 E h U U

Page 4:10

Chapter 4

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 5
Determination of stress intensity factor KI
Problems with solutions Stress intensity factor 5/1.
/2
y x

2h

/2

<< h

An infinitely long strip, height 2h, is split along half its length, see the figure (the strip thus has a semi-infinitely long crack in it). The upper and lower boundary of the strip have been given a prescribed displacement /2 each. Determine the stress intensity factor KI for this configuration. Assume plane stress at the crack tip.

Solution: The strategy to solve this problem is the following: the stress intensity factor KI can be determined from the energy release rate G. To find G, an energy consideration is made. When the crack grows (say a) the strip to the right of the crack tip becomes a shorter, and the energy in that part (i.e. in a volume a by 2h by strip thickness t) is released. This energy is used to determine G. First the elastic strain energy density u stored in the strip far away from the crack tip is determined. To do this, we need to know the stresses and the strains in the strip. The strains in the strip (far away to the right of the crack tip) are yy = (a,b,c) , xx = 0 , and zz = yy 2h The strain xx is zero because no movements can occur in the x direction far away from the crack tip. Also, all shear strain components are zero because of symmetry. Hookes law gives, with zz = 0,

SOLCHAP5.DOC/2010-01-15/TD

Chapter 5

5:1

yy =

1 = { xx} 2h E yy 1 { yy} E xx and xx = yy

(d) (e)

xx = 0 =

from which the stresses are solved: E yy = 1 2 2h

(f,g)

The strain energy density u (Nm/m3) becomes

1 1 E 2 1 u = ij ij = (xx xx + yy yy ) = 2 2 2 1 2 2h (h)

The energy U in a section of the strip with length a is (far away to the right of the crack tip)
1 E 2 U = u Volume = u 2h a t = 2h a t 2 1 2 2h (i)

where t is the thickness of the strip. The strain energy close to the crack tip is not calculated here, because close to the crack tip the stress field is complicated, and, as will be seen, it is not necessary to calculate the strain energy in this region. The energy release rate G is, at displacement control, dU 1 U (a + da ) U (a ) 1 U (a ) U (a + da ) d = = = (j) G = dA t da t da t da Thus, the energy release when the crack grows a distance a is U = U(a) U(a+a). The energy release rate G then becomes (at displacement control), by use of U from (h),
1 U 1 1 E 2 2h t = G= t a t 2 1 2 2h (k)

Finally, KI is obtained from KI = E G , (E = E at plane stress) giving

KI = EG = h 2 2h 1 E (l)

Answer: The stress intensity factor is KI = E / 2 h (1 2)

5:2

Chapter 5

5/2.
P

Use the compliance method to determine the energy release rate G and from that, determine the stress intensity factor KI for a double 2 h cantilever beam (DCB) specimen. Crack length is a, thickness is b and specimen height is 2h, b where a >> b and a >> h. The material is linear elastic with Youngs modulus E. Investigate the two loading cases:

(a) Load control, with the loading force P prescribed. (b) Displacement control, with displacement prescribed. Assume plane stress conditions. Solution: The force versus displacement relationship for a cantilever beam gives the displacement / 2 of the beam end due to the loading force P as P a3 2 P a3 = giving = (a) 2 3 EI 3 EI The compliance C of the specimen is 1 2 P a 3 2a 3 dC 2 a 2 C= = = giving = (b) P P 3 EI 3 EI da EI (a) The energy release rate G due to a crack extension is, for prescribed load P, P 2 dC P 2 2 a 2 G= = (c) 2 b da 2 b EI In the case of plane stress at the crack tip, one obtains
P 2 a 2 P a 2 3 = (d) bI bh h It is seen that the stress intensity factor KI increases as the crack grows, implying that unstable crack growth will be expected (at least if Kc does not depend on crack length a). KI2 G= G E = giving KI = E

(b) The energy release rate G due to a crack extension is, for prescribed displacement , P 2 dC P 2 2 a 2 = G= 2 b da 2 b EI
Using P= 3 EI 2 a3

(e) (f)

Chapter 5

5:3

one obtains

a 2 3 EI 9 EI 2 = G= b EI 2 a 3 4 b a4
2

(g)

In the case of plane stress at the crack tip, one obtains

KI2 G E = giving KI = G= E

Here it is seen that the stress intensity factor KI decreases as the crack grows, implying that stable crack growth will be expected (at least if Kc does not depend on crack length a). Answer: Stress intensity factors are (a) KI = 2 3Pa / bh h , and (b) 2 KI = 3 Eh h / 4a . 5/3.
M

9 E 2 I 2 3 E h h = 3 E h h = 4 2 2 4b a 4a 12 2 a

(h)

2h
a M b

Use the compliance method to determine the energy release rate G for a double cantilever beam (DCB) specimen. Crack length is a, thickness is b and specimen height is 2h, where a >> b and a >> h. The material is linear elastic with Youngs modulus E. The specimen is loaded with bending moments M at the beam ends.

Solution: The (generalised) force versus (generalised) displacement relationship (here this becomes moment versus rotation) of a cantilever beam M gives the rotation angle of the beam end due a to the bending moment M. One obtains Ma 2M a = giving = 2 = (a) EI EI The compliance C of the specimen is 2a dC 2 C= = giving = (b) M EI da EI The energy release rate G due to a crack extension is, for prescribed load M,
12M 2 M 2 dC M 2 2 = = G= 2 b da 2 b EI E b 2 h 3 (c)

Answer: The energy release rate G is G = 12M2 / Eb2h3.

5:4 Chapter 5

5/4.
P

2h
P a a

Use the compliance method to determine the energy release rate G for a beam with a long central crack, see figure. The crack length is 2a, beam thickness is b and beam height is 2h, where a >> b and a >> h. The material is linear elastic with modulus of elasticity E. The beam is loaded symmetrically with two opposite forces P at the beam centre. The crack is assumed to grow symmetrically.

Solution: The force versus displacement relationship of a fixed-fixed beam (length 2a) gives the displacement / 2 of the beam centre due to the loading force P. One obtains P (2a )3 1 1 P a3 = giving = (a) 2 3 EI 8 8 12 EI The compliance C of the specimen is a3 dC a2 C= = giving = (b) P 12 EI da 4 EI Note that, due to symmetry the crack is assumed to grow in both directions, which gives (in d/dA) that dA = d(2ab) = 2bda (A is crack area) (or, it can be seen as one crack with width 2b, i.e. two crack fronts of length b each). The energy release rate G due to a crack extension in both directions then is, for a prescribed force P, 3 P2 a2 P 2 dC P 2 a 2 (c) = = G= 2 (2b ) da 4 b 4 EI 4 E b 2 h 3 Answer: The energy release rate G is G = 3P2a2 / 4Eb2h3.

Chapter 5

5:5

5/5.
q

2h

v(x)

x q

Use the potential energy to determine the energy release rate G for a beam with a long central crack, see figure. The crack length is 2a, beam thickness is b and beam height is 2h, where a >> b and a >> h. The material is linear elastic with modulus of elasticity E. The beam is loaded symmetrically with two opposite distributed loads q (q in N/m) at beam upper and lower surfaces.

The deflection of a clamped-clamped beam loaded with a distributed load q is, for 0 x 2a, q v (x ) = (x 4 4ax 3 + 4a 2x 2) 24 EI
1 d (a) 2b da The crack front length (the extension of the crack tip in the z direction) is 2b, because here there are two crack tips and each crack font has length b. G =

Solution: The energy release rate G for the beam is

The potential energy is, see equation (4.4) in the textbook, q v dS = U W = u dV Vol boundary
1 1 = q v dS q v dS = q v dS 2 2 boundary boundary boundary (b)

where U = W / 2 has been used (the material is linear elastic). Expression (b) 2a gives 1 2a q v (x ) dx + q ( v (x )) dx (c) = 2 0 0 where the first term refers to the upper surface of the beam (loaded upwards) and the second term refers to the lower surface (loaded downwards). Insert v(x) into (c) and integrate. It gives 2a 1 q2 1 4a 5 (2a ) = q v (x ) dx = (2a )4 + 4a 2 (2a )3 24EI 5 4 3 0
2 q2 a5 = 45 EI (d)

5:6

Chapter 5

Finally the energy release rate G can be calculated: 1 d 1 ( 2 q 2) 4 1 q 2 a 4 4 q 2 a 4 G = = 5a = = 2b da 2b 45 EI 9 b EI 3 E b 2h 3 Answer: Energy release rate G is G = 4q2a4 / 3Eb2h3. Reciprocity relation 5/6.
M

(e)

2h
a M b

Determine the stress intensity factor for the structure in Problem 5/3 by use of the reciprocity relation. Use Problem 5/2 as the known Case (1).

Solution: Use equation (5.10) in the textbook to determine the stress intensity factor. One has ui(1) 4 (2) (2) KI = Ti d (5.10) a ( + 1) KI(1) The stress intensity factor in Problem 5/2 is known. Thus, the factor KI(1) to use in the right hand side of (5.10) is (from Problem 5/2)
KI(1) = P a 2 3 bh h (a)

From loading case (1) (Problem 5/2) one needs to know also the (generalized) displacement at the point where the load is going to be applied in the second loading case. This means that one needs to know the rotation angle in Case (1) (due to load P) at the end of the cantilever beam (where the load M is applied in Case (2)). One obtains Pa 2 (b) = 2EI Also, d/da is needed. One obtains
d 2Pa Pa = = da 2EI EI (c)

In Problem 5/3 the load, i.e. the bending moment M, is applied in two discrete points. This means that the intergration in (5.10) along the perimeter can be discretized. Note that the integral in (5.10) has dimension N/m. In a general case one has Ti(2) (N/m2) times dui(1)/da (m/m) times d (m), giving (N/m2) (m/m) m = N/m. When discretizing the integral, Ti(2)d will have the dimension N/m. In case Ti(2) is a stress (N/m2), Ti(2)d becomes force per unit width (width in the z direction) of the structure.
Chapter 5 5:7

Here, when the (generalized) displacement is an angle , one obtains the dimension of d/da as 1/m. The expression Ti(2)d then must have dimension N (unit of force) to give N/m when multiplying Ti(2)d with d/da. This implies that the loading moment M (given in Nm in the problem) should be applied as moment per unit width (in the z direction) when the moment is entered into (5.10). Thus, the discretized version of (5.10) becomes, in this case, ui(1) 4 M Pa 4 (2) (2) KI = T 2 d = (d) i a b EI ( + 1) KI(1) ( + 1) KI(1) Here, at the end of the expression one has d/da from (c), next to that the bending moment per unit width M / b, and in front of M / b the factor 2, which is there because there are two bending moments. Using that = E / 2(1 + ) (the shear modulus), and from equatio (4.34) that = (3 ) / (1 + ) for plane stress conditions at the crack tip, one obtains
KI(2) = 4E (1 + ) bh h 2 M Pa = 2 3 M 2(1 + ) 4 b EI bh Pa 2 3 h (e)

From Problem 5/3 is obtained

EG KI(1) = =

which is in agreement with (e).

E 12M 2 2 3 M = E b 2h 3 bh h KI =

(f)

Answer: The stress intensity factor is, cf. answer to Problem 5/3,

2 3 M bh h

5:8

Chapter 5

Determine the stress intensity factor for the structure in the figure by use of the reciprocity relation. Use Problem 5/2 as the known Case P (1). Case (1) 2 h Investigate some different loading cases of the x structure (structure and loadings are symb metric): a P (a) a general loading q(x) (N/m) along part of the beams, i.e., the loading is q(x) for 0 < x < q ( x) a (note that x = 0 at the free surface and x = a Case (2) at the crack tip), x (b) the loading q(x) is constant, q(x) = q0, along part of the beams, i.e., the loading is q0 for 0 < x q ( x) < a, a (c) what happens in Case (b) above if the factor becomes small ( << 1) but the force q0a at a the beam end is a constant, say q0 a = Q? Assume plane stress conditions, and compare with Problem 5/2. Solution: Use equation (5.10) in the textbook to determine the stress intensity factor. One has ui(1) 4 (2) (2) KI = Ti d (5.10) a ( + 1) KI(1) The stress intensity factor in Problem 5/2 is known. Thus, the factor KI(1) to use in the right hand side of (5.10) is (from Problem 5/2)
P a 2 3 (a) bh h From loading case (1) (Problem 5/2) one needs to know also the (generalized) displacement at the points where the load is going to be applied in the second loading case. This means that one needs to know the deflection in Case (1) (due to load P) of the cantilever beam where the load q(x) is applied in Case (2). One obtains, with ui(1)(x) = v(x), Pa 3 3x x 3 v (x ) = 2 (b) + 6EI a a3 KI(1) =

5/7.

Also, v /a is needed. One obtains v (x ) P = (2 3a 2 3x 2a + 0 ) a 6EI

Chapter 5

(c)
5:9

Equation (5.10) now gives a 4 q (x ) P (6a 2 6xa ) dx KI(2) = 2 b 6EI 0 ( + 1) KI(1)

(d)

where = E / 2(1 + ) is the shear modulus, = (3 ) / (1 + ) (plane stress), and the factor 2 is there due to the two cantilever beams (integration along two beams). In (5.10) T(2) is stress. Dividing q(x) (N/m) by the thickness b gives the stress on the boundary. (b) Enter q(x) = q0 into (d). It gives a q a 4 P 0 (2) KI = 2 (a x ) dx (1) b EI 0 ( + 1) KI
= q0 a P 4 1 2 2 a a ( a ) 2 ( + 1) KI(1) b EI
(2) I

(e) (f)

Thus,
K

P a 2 q0 a 8 = 1 (1) b 2 ( + 1) KI EI

(c) Enter q0a = Q into (f), and then let go to zero. It gives
KI(2) = = P a2 Q 8 8 b h h P a 2 Q = ( + 1) KI(1) EI b ( + 1) P a 2 3 EI b (g)

8E (1 + ) b h h a 12 Q = Qa 2 3 2(1 + ) 4 2 3 Ebh 3 b bh h as obtained in Case (1) (at plane stress), see expression (a) above.

Answer: (a)

KI(2) =

a 4 q (x ) Pa (a x ) dx 2 b EI 0 ( + 1) KI(1)

(d) (f)

(b) Entering q(x) = q0 into (d) gives

KI(2) =

P a 2 q0 a 8 1 b 2 ( + 1) KI(1) EI KI(2) = Qa 2 3 bh h

(c) Enter Q = q0a and = 0 into (f). It gives as in Case (1) (at plane stress), cf. expression (a).

5:10

Chapter 5

Superposition 5/8. Demonstrate (no calculations needed) why the two loading cases in the figures have the same stress intensity factor KI, cf. (j) and (k) in Section 5.2.3 in the textbook. The plates are large. In Case (2) the stress is a compressive stress acting on the two crack surfaces (i.e., is a pressure loading the crack surfaces).

Case (1)

Case (2)

2a

2a

Figure Loading case (1) with remote stress and loading case (2) with same stress acting on the crack surfaces Solution: In Case (2), replace the compressive stress on the crack surfaces with a tension stress . Call this loading case for Case (2b). Superimpose loading case (2b) on Case (1). It gives the same stress distribution in the plate (in the plane of the crack) as in a plate with no crack at all, see figure below. This implies that the stress intensity factors in the two cases must cancel each other, and therefore they must be the same.

Chapter 5

5:11

No crack

Case (1)

Case (2b)

2a

2a

2a

Figure Loading case (1) with remote stress and loading case (2b) with tension stress acting on the crack surfaces Or, doing it in anther way: Start with a plate without crack, see the leftmost part of the figure. The stress in the left plate is uniaxial (value ) and evenly distributed all over the plate. Create loading case (1) by cutting a crack in the first plate. It means that in the second plate the stress cannot be transferred from the upper crack surface to the lower crack surface (as was done in the first plate). This gives the loading case (1) with a certain stress intensity factor KI(1). Next, the stress that could not be transferred over the crack in loading case (1) is applied on the crack surfaces of another plate. This gives loading case (2b), see the figure. The loading case (2b) gives a stress intensity factor KI(2b). Superimposing Cases (1) and (2b) gives the first plate with no crack. But the first plate has no stress singularity, so one must have
KI(1) + KI(2b) = 0 giving KI(1) = KI(2b) (a)

Finally, changing the sign of the stress in loading case (2b) gives back the original loading case (2), and one has
KI(1) = KI(2) (b)

Answer: Replace the compressive stress in Case (2) by a tensile stress (of magnitude ) on the crack surfaces. This gives the Case (2b). Superimpose Case (1) and the new Case (2b). It gives a plate with a homogeneous stress field throughout the plate, without stress concentration. Thus, the two stress intensity factors cancel each other, and it follows that Case (1) and Case (2) must have the same stress intensity factor.

5:12

Chapter 5

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 6
Crack propagation under cyclic loading
Problems with solutions Crack propagation at cyclic loading, Paris law 6/1. Crack-like flaws may develop in most materials. These flaws should be treated as if they were cracks. Therefore, assume that the flaws are penny-shaped (the crack looks like a coin with a sharp egde). Determine the largest flaw allowable with respect to fatigue crack propagation in the following materials. The loading is cyclic and the stress range is 0.5Y (and Y is the yield limit of the material). (a) A low-strength steel with Kth = 7 MN/m3/2 and Y = 300 MPa, and (b) a high-strength steel with Kth = 4 MN/m3/2 and Y = 1500 MPa. Solution: A penny-shaped crack in the interior of the material will give the stress intensity factor KI = 0 a f8(a ). Case 8 in Appendix 3 of the textbook gives (terms containing a / t can be neglected here so that f8 = 0.637 is obtained)
KI = 0.637 0 a (a)

(a) No crack propagation is allowed. Then the stress intensity range KI must be less than (or equal to) the threshold value Kth. Using Kth = 7 MN/m3/2 and 0 = Y / 2 = 150 MPa, one obtains
3/2 KI = 0.637 150 a = 7 MN/m

(b)

giving a = 1.7 mm. (b) Similarly, using Kth = 4 MN/m3/2 and 0 = Y / 2 = 750 MPa, one obtains
KI = 0.637 750 a = 4 MN/m3 / 2 (c)

giving a = 0.0223 mm.

FMCHAP6P.DOC/2010-01-15/TD

Chapter 6

Page 6:1

It is concluded that the high-strength steel is much more sensitive to cracks than the low-strength steel. (One reason is that the high-strength steel is loaded at higher stress.) Answer: Critical crack length is (a) a = 1.7 mm, and (b) a = 0.022 mm. 6/2.
P

A double cantilever beam (DCB) specimen is subjected to a pulsating loading. The crack length is a, thickness is b and specimen height is 2h, where a >> b and a >> h. The material is 2 h linear elastic with Youngs modulus E. Use Paris law da b = C (KI )n dN to determine the number of loading cycles required to make the crack grow from its initial length ai to a final length a, if

(a) the load is prescribed; the load varies between 0 and Pmax, (b) the displacement is prescribed; the displacement varies between 0 and vmax (in the figure one has = 2vmax). Solution: (a) The stress intensity range becomes (p = Pmax /b (N/m) is the load range)
h h Entering (a) into Paris law gives KI = 2 3 p a= 2 3 Pmax b h h a
n 2 3 Pmax a h b h

(a) (b) (c)

da = C (KI)n = C dN da =C an
n

Separating (b) gives

n 2 3 Pmax dN b h h

Integrating (c) and solving for N give

b h 1 h 1 n (a ai1 n ) N= C (1 n ) 2 3 Pmax

(n 1)

(d)

Page 6:2

Chapter 6

(b) The stress intensity range becomes (here vmax = /2 is the range of the displacement) 3 E h 2 vmax 1 KI = (e) a2 2 h n Entering (e) into Paris law gives 2 3 E h v da 1 max = C (KI)n = C (f) 2 dN a 2 h Separating (f) gives a N 3 E h 2 vmax n a 2n da = C (g) dN ai 0 2 h Integrating (g) and solving for N give
2 1 h 2n + 1 (a ai2n + 1) N= 2 C (2n + 1) 3 E h vmax
n

(h)

Answer: Number of cycles required is, respectively, n (a) b h 1 h 1 n 1 n (a N= ai ) C (1 n ) 2 3 Pmax (b)

n

(n 1)

2 1 h 2n + 1 2n + 1 (a ai ) N= C (2n + 1) 3 E h 2 vmax

6/3. A structure is subjected to a pulsating loading, i.e.. min = 0. Embedded circular (penny-shaped) cracks are discovered in the structure. The crack diameters are 2a0, which is much smaller than the thickness of the material. The structure is made of a material that follows the crack propagation law da = C (KI )n dN (a) Determine the largest (remote) stress range 0 the structure may be subjected to if crack propagation is not allowed. (b) Determine the maximum stress range that may be allowed if the structure is loaded by N = 10 000 cycles. During the loading the crack is not allowed to grow so large that it becomes unstable. Therefore, choose safety factor s = 1.5. Numerical data: crack length a0 = 0.0005 m, Y = 620 MPa, Kth = 2 MN/m3/2, KIc = 36 MN/m3/2, n = 4.0 and C = 1.1210 11 m7/MN4.

Chapter 6

Page 6:3

Solution: (a) No crack propagation If no crack propagation is allowed, the stress intensity range KI must be less than or equal to the threshold value Kth. Assume that there exists a crack with the most unfavourable direction. The stress intensity factor KI then is (Case 8 in Appendix 3 in the textbook, with a << t) a KI = 0 a f8 , 0 (a) c where a = c, which gives f8(1,0) = 0.637. Thus
a KI = 0.637 0 (b) (c)

The stress intensity range KI becomes (0 is stress range)

a KI = 0.637 0

No crack propagation is expected if KI Kth, which gives

0.637 0 a Kth (d)

Solving for 0 gives Kth 2 106 0 = = 79 106 N/m2 = 79 MPa 0.637 a 0.637 0.0005 Thus, if 0 79 MPa no crack growth is expected.

(e)

(b) Crack propagation After N = 104 loading cycles the crack should give a stress intensity factor KImax = KIc / s. Determine the critical crack length ac giving this stress intensity factor. Assume that the crack stays circular (penny-shaped) during the propagation. One then has K = 0.637 a (f)
I 0

Fracture will occur (if LEFM is valid) when KIc KImax = s Equations (f) and (g) give
KIc KIc 2 1 36 2 452 1 = 2 0.6370 = ac = s giving ac = 0 0.637 0 s 0.637 0 1.5

(g)

(h)

where, here, 0 (in MPa) equals the stress range 0 (in MPa) because min = 0 (load is pulsating).
Page 6:4 Chapter 6

Now the initial crack length a0 and the final crack length ac (expressed in the unknown stress range 0) are known. During N cycles the crack will grow from a0 to ac. The crack propagation law (Paris law) gives da n a ) (i) = C (KI)n = C (0.637 0 dN ac which gives N da n = C ( 0.637 ) dN (j) 0 n a0 ( a ) 0 1 n /2 Evaluation of the integrals gives ac1 n /2 a0 n ) N (k) = C (0.637 0 1 n /2 Entering the values of a0, ac, C, n and N into (k) gives
(0.5 10 3 ) 4 ) 104 = 1.12 10 11 (0.637 0 1 Solving for 0 gives 0 = 324 MPa = 0. (452/(0)2)
1 1

(l)

Finally, LEFM was used when ac was determined. Is LEFM valid? The stress 0 = 324 MPa (= max) gives the critical crack length 452 = 0.0043 m ac = 3242 to be compared with

(m)

KIc 2 36 2 (n) = 0.0037 m 2.5 = 2.5 1.5 620 s Y Thus, ac is larger than 0.0037 m so LEFM may be used (t and W a are assumed large enough).

Note that only the crack length ac at fracture is of interest here. The fracture criterion KImax = KIc / s was used to determine ac. Only then LEFM must be valid, not during the crack propagation phase. Answer: The stress range should be less than (a) 0 = 79 MPa, (b) 0 = 324 MPa.

Chapter 6

Page 6:5

6/4. A large plate of thickness t contains an embedded elliptical crack, see figure. Determine the cyclic life (the number of cycles to failure) of the plate if (a) the load varies between 0 and , (b) the load varies between 0.5 and . The influence of the mean value of the stress intensity factor is disregarded. The crack is assumed to keep its form during the crack propagation (i.e., ratio a / c is constant during the crack propagation).

2a
t

2c

The material follows Paris law for crack propagation. Use safety factor s = 1.4. Numerical data: crack length a0 = 0.001 m and c0 = 0.002 m, thickness t = 0.10 m, yield limit Y = 1200 MPa, threshold value Kth < 6 MN/m3/2, fracture toughness KIc = 60 MN/m3/2, and crack propagation parameters n = 4.0 and C = 510 13 m7/MN4, = 400 MPa. Solution: For an embedded elliptical crack the stress intensity factor is (Case 8 in Appendix 3; a << t implies that terms containing a / t may be neglected) a 1 KI = a f8 , 0 = 0.826 (a) , 0 = a f8 a c 2 The stress intensity range KI becomes
KI = 0.826 a (b)

The initial crack length a0 = 0.001 m gives the stress intensity range Case (a):
0.001 KI = 0.826 a = 0.826 400 = 18.5 MN/m 3 / 2 > Kth Case (b): KI = 0.826 0.001 a = 0.826 200 = 9.26 MN/m 3 / 2 > Kth (d) (c)

Thus, the crack will grow in both cases.

Page 6:6

Chapter 6

Next, determine the critical crack length ac. Failure will occur (if linear elastic fracture mechanics theory, LEFM, may by used) when KIc 60 KImax = ac = giving 0.826 400 (e) s 1.4 which gives ac = 0.0054 m at failure (or, to be exact, the crack will start to grow rapidly at this crack length and the failure will come after a few more cycles). Check the conditions for LEFM. One has
KIc 2 60 2 (f) = 0.0032 m = 2.5 2.5 1.4 1200 s Y Thus, ac > 0.0032 m and LEFM can be used (also t and W a are large enough).

Paris law now gives

da n a ) = C (KI)n = C (0.826 0 dN (g)

Separating and integrating give

N cr 1 n C (0.826 d a = ) dN 0 n a0 ( 0 a ) a

(h) (i)

Thus

N=

1n /2 1n /2 a0 acr

(1 n / 2) C (0.826 0 )

Case (a) gives:

N =
(a)

0.0054 1 0.001 1 ( 1) 5 10
13

(0.826 400 )

= 13 855 cycles

(j)

Case (b) gives:

( 1) 5 10 13 (0.826 200 ) Thus N(b) = 24 N(a) = 221 600 cycles. N
(b)

0.0054 1 0.001 1
4

= 24 N (a) = 221 680 cycles

(k)

Answer: Cyclic life to failure is, approximately, (a) N = 13 850 cycles, and (b) N = 221 600 cycles (cf. Problem 6/9).

Chapter 6

Page 6:7

6/5.

2c

A large plate of thickness t contains an elliptical surface crack, see figure. The load is cyclic and it varies between 0 and . (a) Determine the relationship between the remote stress and the number of loading cycles N for crack propagation to depths a = 5, 10, 15 and 20 mm. (b) Determine also the stress level that may give unstable crack growth at the different crack depths. The crack is assumed to keep its form during the crack propagation. The material follows Paris crack propagation law. Use safety factor s = 1.5.

Numerical data: crack length a0 = 0.002 m, c0 = 0.004 m, t = 0.10 m, Y = 1200 MPa, Kth = 6 MN/m3/2, KIc = 70 MN/m3/2, n = 3.75, C = 9.2210 12 m6.625/MN3.75. Solution: Here the stress intensity range becomes (Case 7 in Appendix 3 in the textbook, terms containing a / t may be neglected here) 1 KI = (a) , 0 = 0.896 a f7 a 2 Paris law then becomes da 3.75 a) (b) = C (KI)n = C (0.896 dN Separating and integrating give
N final da 3.75 = C ( 0.896 ) dN 3.75 ainitial ( 0 a ) a

(c) (d)

Thus or, rewritten,

1 1 1 3.75 N 0.875 0.875 = () 3.75 0.875 a a final initial C (0.896 ) 1 3.75 log + logN = log 1 1 1 0.875 0.875 3.75 0.875 a a final initial C (0.896 )

(e)

which gives, for a = afinal = 5, 10, 15, and 20 mm, respectively, 3.75 log + logN = 12.44, 12.57, 12.62, and 12.64

(f)

Page 6:8

Chapter 6

(b) The critical stress at the different crack lengths is, from (a), (and if LEFM can be used) 1 KImax = ( KI = ) 0.896 (g) a = 1.5 KIc = 46.7 MN/m3 / 2 Check for LEFM. One has KIc 2 2.5 (h) = 0.00378 m s Y implying that one may use LEFM. With (g), the four crack lengths 5, 10, 15, and 20 mm give, respectively, = 416, 294, 240, and 208 MPa (i) Answer: (a) 3.75log + log N = 12.44, 12.57, 12.62, and 12.64 for a = 5, 10, 15, and 20 mm, respectively, and (b) = 416, 294, 240, and 208 MPa, respectively.

6/6.
l P

10 l
a

A beam with an edge crack is loaded with two forces P = P0 sint, see figure. Determine a t lower limit of the cyclic life of the beam h (assume that no other cracks are present). The material follows Paris crack propagation law. Use safety factor s = 1.5.

Numerical data: crack depth a0 = 0.005 m, thickness t = 0.01 m, height h = 0.10 m, length l = 0.10 m, yield limit Y = 1400 MPa, threshold value Kth = 4 MN/m3/2, fracture toughness KIc = 70 MN/m3/2, and crack propagation parameters n = 4 and C = 110 13 m7/MN4. Load amplitude is P0 = 0.042 MN. Solution: Between the two forces P, the bending moment in the beam is constant and it is Pl. The maximum value of the bending moment is P0 l and the minimum value contributing to the stress range is zero (it is assumed that the negative part of the moment will not contribute to the crack growth). Thus, the range of the bending moment contributing to the stress range is P0 l. The stress intensity factor becomes (Case 6 in Appendix 3 in the textbook) a 6Pl KI = a f (a) 6 W t W2 For crack length a = a0 = 0.005 m one has a / W = a0 / h = 1/20 = 0.05, which gives f6(0.05) = 1.08.
Chapter 6 Page 6:9

Determine the crack length ac at failure. One obtains 6 P0 l ac KIc = KI = a f c 6 W s t W2

(b)

Here f6(ac / W) is unknown so one has to try (to guess) a value of f6. Try f6(ac / W) = 1.06 to see what happens (we see from the diagram that f6 decreases in the region where a / W = 0.05). It gives, for stress and fracture toughness expressed in MPa, 6 0.042 0.1 70 a 1.06 = (c) KI = c 1.5 0.01 0.102 Solving for ac gives ac = 0.0097 m. This value of ac gives f6(ac / W) = f6(0.097) = 1.06, which is the same as the value assumed. Therefore, the critical crack length ac = 0.0097 is selected for the following calculations. Paris law now gives
4 6 P0 l da a n (d) = C (KI) = C a f6 dN W t W2 Again, we have the problem of not knowing the value of the function f6 = f6(a) when the crack grows, i.e. when a changes. From the above calculations (and from the form of the function f6 seen in the diagram) it is concluded that f6 varies between 1.08 and 1.06 in the crack length interval obtained here. Therefore, to be on the safe side, select f6 = 1.08 for the whole interval (this is a conservative estimation, as it will give a too large crack propagation rate). Now equation (d) will give (with stresses in MPa, see the dimension of the constant C) 6 P0 l 4 da =C (e) 1.08 dN a2 t W2 0.0097 6 P0 l 4 da =N C (f) 1.08 0.005 a2 t W2 from which N = 17 900 is solved. As we have used a too large value of f6 this will be a lower limit of the cyclic life of the structure. The cyclic life is expected to be larger than the life calculated. Thus, it is expected that N will be slightly more than 17 900 cycles. (An upper limit of the fatigue life is estimated if f6 = 1.06 is used. Then one obtains, see equation (f), Nupper = 17 900 (1.08/1.06)4 = 19 290 cycles.)

or

Answer: The fatigue life is expected to be slightly more than N = 17 900 cycles (but less than 19290 cycles). Critical crack length is acrit = 0.0097 m.
Page 6:10 Chapter 6

6/7. In a large plate of a ships hull two cracks of approximately the same size were discovered at a hole, see figure. The plate is loaded in cyclic tension with min = 0.2 and max = = 60 MPa. The loading frequency is 6 cycles per minute. Determine the remaining life (in hours) of the plate. It is assumed that the crack propagation is symmetric with respect to the hole.

a 2r a

The material follows Paris law for crack propagation. Use safety factor s = 1.5. The geometry function f used to determine the stress intensity factor may in this example, and for the crack length used here, be approximated to
a a 1/6 f = 1.42 r r (Thus, this is an approximation of function f3 in Appendix 3 of the textbook.)

Numerical data: r = a0 = t = 0.08 m, Y = 600 MPa, KIc = 90 MN/m3/2, n = 3, C = 110 11 m11/2/MN3. Solution: In this example, the stress intensity factor may be written, approximately, a 1.42 KI = 0 a f = 0 a = 1.42 0 a 1 / 3r 1 / 6 (a) 1/6 r (a /r ) (Note that the function f given here is an approximation of the function f3 given in Appendix 3 in the textbook.) Determine the critical crack length acr. Equation (a) gives, by use of KImax = KIc / s,
KIc 3 3 90 1 = 0.22175 m a = acr = = 60 0.081 / 6 s 1.42 r 1 / 6 1.5 1.42 (b)

May linear elastic fracture mechanics (LEFM) be used here? One has
KIc 2 (c) 2.5 = 0.025 m s Y Yes, all three measures a, t, and W a are larger than 0.025 m, so LEFM may be used. The critical crack length is acr as calculated in (b).

Chapter 6

Page 6:11

The stress intensity range becomes = max min = 0.8, which gives
KI = 1.42 0.8 a 1 / 3 r 1 / 6 (d) (e) (f) (g)

Paris law gives Thus giving

a

da 3 0.8 a 1 / 3 r 1 / 6 ) = C (KI)n = C (1.42 dN

N cr 3 da = C (1.42 0.8 r 1 / 6 ) dN a0 a 0

N=

acr ln = 204 400 cycles 3 C (1.42 0.8 r 1 / 6 ) a0 1

Remaining fatigue life is t = N / (6 60) = 568 h. Answer: Fatigue life N = 204 000 cycles gives that the remaining life is expected to be 568 hours. 6/8. A large plate of thickness t contains a halfelliptical surface crack, see figure. The load t consists of a repeated sequence of two cycles as shown in the figure. Determine the expected 2c remaining life of the plate expressed in number of sequences to failure. It is assumed that the crack keeps its form during the crack propagation. The material follows Paris law. time Use safety factor s = 1.4.

2 1

Numerical data: a0 = 0.002 m, c0 = 0.004 m, t = 0.100 m, Y = 1200 MPa, KIc = 70 MN/m3/2, Kth = 6 MN/m3/2, n = 4, C = 110 13 m7/MN4, 1 = 200 MPa and 2 = 400 MPa . Solution: A half-elliptical surface crack (Case 7 in Appendix 3 in the textbook) gives the stress intensity factor, for a / c = 0.5, a KI = 0 a f7 , 0 = 0 a 0.896 = 0.90 0 a (a) c The range of the stress intensity factor becomes
a KI = 0.90 0 (b)

Page 6:12

Chapter 6

Will the lower stress level 1 = 200 MPa contribute to the crack growth? Let a = a0 = 0.002 m and 0 = 1 = 200 MPa. One then obtains from (b)
KI = 0.90 200 0.002 = 14.3 MN/m3 / 2 (c)

The threshold value is Kth = 6 MN/m3/2, giving KI > Kth. This implies that the crack will propagate at the lower stress cycle 1, and thereby also at the higher stress level 2. Determine next the crack length acrit at fracture. Assume that linear elastic fracture mechanics (LEFM) is valid. Fracture will then occur when KImax = KIc / s. It gives KIc KImax = 0.90 0 acrit = (d) s Fracture will occur when the highest stress is loading the structure, i.e. when the stress is 2. For 0 = 2 one obtains from (d) (if linear elastic fracture mechanics is valid)
KIc 2 1 1 70 2 acrit = = 0.0061 m = s 0.90 2 1.4 0.90 400 (e)

Is linear elastic fracture mechanics valid?

KIc 2 70 2 2.5 (f) = 0.0043 m = 2.5 1.4 1200 s Y which is less than acrit, t, and W a. Thus, linear elastic fracture mechanics can be used.

It is now known that the crack will grow from a0 to acrit, and both cycles in the sequence will contribute to the crack propagation. Per sequence of two cycles, the crack propagation will be
da n = C (KIi )n = C (0.90 0i a ) dNs = C (0.90 1 a ) + (0.90 2 a )

(g)

Note that Ns stands for the number of sequences (not number of cycles). Separation of variables gives
crit s da n n = C (0.90 1 ) + (0.90 2 ) 0 dNs n a0 ( a )

(h)

Chapter 6

Page 6:13

which gives

n n 1 n /2 1 n /2 n acrit a0 = 1 C (0.90 ) { n 1 + 2 } Ns 2

(i)

Entering acrit, a0, 1, 2, and the material parameters C and n in (i), and then solving for Ns, give (j) Ns = 19 080 sequences Answer: Expected fatigue life is Ns = 19 000 sequences, giving N = 38 000 cycles to failure. 6/9. A large plate of thickness t contains an embedded elliptical crack, see figure. Determine the cyclic life (the number of cycles to failure) of the plate if the load varies between 0.5 and . The influence of the mean value of the stress intensity factor has to be taken into account. The crack is assumed to keep its form during the crack propagation (i.e., ratio a / c is constant during the crack growth). (This is Problem 6/4(b) once again, but now the influence of the mean value is taken into account.)

2a
t

2c

The material follows Paris law for crack propagation. Use safety factor s = 1.4. Numerical data: crack length a0 = 0.001 m, c0 = 0.002 m, t = 0.10 m, Y = 1200 MPa, Kth < 6 MN/m3/2, KIc = 60 MN/m3/2, n = 4.0 and C1 = 510 13 m7/MN4, = 0.7, = 400 MPa (it is assumed that the material parameters C and n in Paris law are the same here as in Problem 6/4). Solution: Following the solution to Problem 6/4, one obtains the stress intensity factor for an embedded elliptical crack as a a 1 KI = a f8 , 0 = 0.826 (a) , = a f8 a c t 2 The stress intensity range KI becomes
KI = 0.826 a
Page 6:14 Chapter 6

(b)

The initial crack length a0 = 0.001 m gives the stress intensity range
KI = 0.826 0.001 a = 0.826 200 = 9.26 MN/m 3 / 2 > Kth (c)

Thus, the crack will grow at the stress range given. Determine the critical crack length ac. Failure will occur (if linear elastic fracture mechanics theory, LEFM, may by used) when KIc 60 KImax = ac = (d) giving 0.826 400 MN/m 3 / 2 s 1.4 which gives ac = 0.0054 m at failure (or, to be exact, the crack will start to grow rapidly at this crack length, and the failure will come after a very limited number of loading cycles after this crack length has been reached). Check the conditions for LEFM. One has
KIc 2 60 2 (e) = 0.0032 m = 2.5 2.5 1.4 1200 s Y Thus, ac > 0.0032 m and LEFM can be used (also t and W a are large enough).

Paris law, with correction for the mean stress value according to Walker, now gives C1 da n (KI)n = C (0.826 0 a ) (f) = n (1 ) dN (1 R ) where stress ratio R is R = 0.5. The factor C becomes C = C1 / (1 R )n (1 ) = 1.148710 12 m7/MN4. Separating and integrating give
N cr 1 n d a = C ( 0.826 ) dN (g) 0 n a0 ( 0 a ) 1 n /2 1 n /2 a0 acr a

giving

N=

(1 n /2) C (0.826 0 ) 0.0054 1 0.001 1

(h)

Thus,

N =
(a)

( 1) 1.1487 10 12 (0.826 200 )

= 96 500 cycles

(i)

The expected fatigue life now becomes N = 96 500 cycles, to be compared with N = 221 700 cycles in Problem 6/4. These two numbers differ by the factor (1 R )n (1 ) = 0.435, which means that in this case the mean value plays an important role for the fatigue life of the structure. Answer: Expected fatigue life is N = 96 500 cycles.
Chapter 6 Page 6:15

6/10.
0

thickness t
a

Determine the number of loading cycles required to make the crack in the figure grow by 1 mm. The structure is loaded with stress cycles of constant amplitude. The stress range 0 is 0 = 100 MPa. Use some different initial crack W >> a lengths ai, say ai = 1, 5, 10, 20, 40, 60, and 80 mm.

For crack growth calculations use both the un-modified Paris law: da = C (KI)n dN and the modified version, taking the closeness to the threshold value into account: da = C {(KI)n (Kth)n } dN Compare the results. Assume that the constants C and n are the same in the two formulae: C = 2.510 12 (units in MN and meter) and n = 3.2. The material has the fatigue threshold value Kth = 6 MN/m3/2. Solution: First investigate if crack propagation is obtained for the shortest crack length ai = 1 mm. The intensity range for the component with the edge crack is (Case 5 in Appendix 3 in the textbook, use that W >> a) a KI = 0 (a) = 1.12 0 a f5 a W The stress range 0 = 100 MPa gives, for crack length ai = 1 mm,
a KI = 0 0.001 = 6.28 MN/m / 2 (b) = 1.12 100 a f5 W which is larger than the threshold value Kth = 6 MN/m3/2. It is concluded that the crack will grow for all crack lengths ai.

Paris law gives where KI is obtained from (a). Separating and solving for N give

da = C (KI)n dN

(c)

Page 6:16

Chapter 6

(1.12 100 ) a The modified crack propagation law gives

3.2

N =N

(d)

i 1 = C ai

a + 0.001

da
1.6

(d)

1 i N =N = C ai
(e)

a + 0.001

da (1.12 100 ) a 1.6 (Kth)3.2

3.2

(e)

Numerical solution (for example by use of MATLAB) of (d) and (e) gives results according to the table below. It is noted that only when the stress intensity range is very close to the threshold value there is a large difference between the two results (about one million cycles). A difference was expected, since the crack growth rate tends to zero at the threshold value, whereas the unmodified Paris law predicts a finite crack growth rate also at the threshold value. Already at the crack length 5 mm the difference in crack growth rate between the two formulae is small; only a few per cent. Making the crack grow from 1 mm to 81 mm would, according to the unmodified Paris law, require 1.7310 6 cycles, whereas the modified version predicts 2.8910 6 cycles. The main part of this difference (one million cycles) depends on the slower crack growth the first millimetre. Table: Number of loading cycles required to make the crack grow 1 mm for different initial lengths ai of the crack. The fatigue life N(d) is calculated with the un-modified Paris law whereas N(e) is calculated with the modified crack propagation law given in the problem. It is seen that the two formulae give different results only when the stress intensity range KI(ai) is very close to the threshold value Kth (here Kth = 6 MN/m3/2). ai (m) 0.001 0.005 0.010 0.020 0.040 3 001 3 008 39.7 0.060 1 579 1 581 48.6 0.080 1 000 1 000 56.1

N(d) (cycles) 0.6310 6 73 628 26 056 8 923 N(e) (cycles) 1.6810 6 78 105 26 592 8 985 KI(ai) (MN/m3/2) 6.27 14.0 19.9 28.1

In the table is also given the stress intensity range KI(ai) for the initial length ai of the crack. It is noted that the fracture toughness KIc of the material must
Chapter 6 Page 6:17

be so large that after reduction with the safety factor s the fracture toughness should be larger than KI(ai + 0.001) = KI(0.081) = 56.5 MN/m3/2. Thus, KIc must be such that KIc > s56.5 MN/m3/2. Answer: The un-modified propagation law gives N = 0.6310 6, 73 600, 26 060, 8 900, 3 000, 1 580, 1000 cycles, and the modified formula gives N = 1.6810 6, 78 100, 26 600, 9 000, 3 000, 1 580, 1000 cycles, respectively.

6/11. A large plate of thickness t contains a halfelliptical surface crack, see figure. The load consists of a repeated sequence of two cycles as t shown in the figure. The crack is assumed to keep its form during the crack propagation. The 2c material follows Paris law. The crack is detected when the crack length a0 is a0 = 0.002 m. It is decided that the crack can be allowed to grow to the length afinal = 0.006 m before the time crack has to be repaired.

2 1

(a) Determine the expected number of cycles (twice the number of sequences) required to make the crack grow from a0 to afinal. Use Paris law to determine the crack propagation cycle by cycle. (b) What number of cycles would be obtained if the root mean square value (i.e. Krms) were used as an equivalent measure of the stress intensity range? Numerical data: crack depth a0 = 0.002 m, c0 = 0.004 m, t = 0.100 m, Y = 1200 MPa, KIc = 70 MN/m3/2, Kth = 6 MN/m3/2, n = 4, C = 110 13 m7/MN4, 1 = 200 MPa and 2 = 400 MPa. Solution: A half-elliptical surface crack (Case 7 in Appendix 3 in the textbook) gives the stress intensity factor (for a / c = 0.5) a KI = 0 a f7 = 0.896 0 a (a) c The range of the stress intensity factor becomes
a KI = 0.896 0
Page 6:18 Chapter 6

(b)

Will the lower stress level 1 = 200 MPa contribute to the crack growth? Let a = a0 = 0.002 m and 0 = 1 = 200 MPa. One obtains
KI = 0.896 200 0.002 = 14.2 MN/m 3 / 2 (c)

The threshold value is Kth = 6 MN/m3/2, giving KI > Kth. This implies that the crack will propagate for the lower stress cycle 1, and thereby also for the higher stress cycle 2. Check the stress intensity factor for the higher stress 2 and crack length afinal = 6 mm. One obtains
KI = 0.896 400 0.006 = 49.2 MN/m 3 / 2 (d)

Thus, at the planned repair of the crack, the stress intensity factor is KI = 49.2 MN/m3/2, which is 70 per cent of the critical value KIc = 70 MN/m3/2 (the safety factor is s = 70 / 49.2 = 1.42). It is given in the problem that the crack will grow from a0 to afinal. Both cycles in the sequence will contribute to the crack growth. Then, per loading sequence the crack growth will be da n = C (KI i )n = C (0.896 0 i a ) dNs
=C (0.896 1 a ) + (0.896 2 a )

(e)

Note that Ns here stands for the number of sequences (not number of cycles). Separation of variables gives
a0
afinal

= C (0.896 1 dNs ) + (0.896 2 ) n 0 ( a )

da

Ns

(f)

which gives
n n 1n /2 1n /2 n a0 = 1 C (0.896 ) { n afinal 1 + 2 } Ns 2 (g)

Entering afinal, a0, 1, 2, and the material parameters C and n in (g), and then solving for Ns, give Ns = 19 265 sequences, giving N = 38 500 cycles (h)

Chapter 6

Page 6:19

(b) Using the rms value as an equivalent stress intensity range, one obtains
Krms =

{ (Ki )2 ni } = 0.896 a ni

2002 + 4002 2 (i)

= 0.896 a 316.23 MN/m 3 / 2 Thus, the equivalent stress range becomes eq(rms) = 316.23 MPa.

Paris law then gives

n n n 1 n /2 1 n /2 afinal a0 = 1 C (0.92 ) N (j) ((rms) ) eq 2 giving N = 52 400 cycles. Thus, the rms value here predicts 52 400 / 38 500 = 1.36 times too many cycles. During the 52 400 38 500 13 900 extra cycles, the crack will grow (far) beyond the 6 mm limit where the crack should have been repaired. Conclusion: The conclusion is that the rms value can not be used for crack propagation calculations.

(c) Extra problem: Use N = 52 400 cycles (26 200 sequences) to see what crack length would have been obtained if Paris law were used as in problem (a). One obtains n n 1 n /2 1 n /2 n a0 = 1 C (0.896 ) { n (k) afinal 1 + 2 } 26 200 2 giving afinal = 0.0214 m = 21.4 mm. This crack length gives the stress intensity factor, at stress 2,
a = 0.896 400 0.0214 = 92.9 MN/m 3 / 2 KI = 0.896 2 (l)

which is much larger than the fracture toughness KIc of the material. The structure would fail long before the 52 400 cycles had been applied. Comment: The importance (influence) of the high amplitude loading cycles (here 2) is diminished when the rms value is used. Instead, use
{ (KIi )n ni } 1 / n KI equivalent = ni

(m)

where n is the factor in Paris law (and ni is number of cycles at each stress level). One then obtains the equivalent stress intensity range

Page 6:20

Chapter 6

K I equivalent

{ (KIi )n ni } 1 / n = = 0.896 a ni
3/2 = 0.896 a 341.5 MN/m

2004 + 4004 1 / 4 2

(n)

This value of the stress intensity range gives n n 1 n /2 1 n /2 a0 = 1 C (0.896 ) { 341.54 } 38 500 afinal 2 which gives afinal = 0.006 m, as it should.

(o)

Answer: (a) 38 500 cycles will propagate the crack to length 6 mm, (b) using the rms value of the stress intensity range, 52 400 cycles are obtained, which is 36 per cent too many cycles. That many cycles would propagate the crack to a length that would be longer than the critical value of the crack length. Thus, failure would occur during the loading if the rms value were used in the calculations.

6/12. The irregular loading sequence of a machine component has been analysed and reduced to the 60 stress range cycles given in the table below. Upon inspection of the component, edge cracks of depth a = 1 mm (or longer) will be detected and eliminated. The maximum permissible crack length in the component is a = 5 mm, see figure. Stress range 0 (MPa) 285 Number of cycles N
0

270 3

200 12

92 44

thickness t
a

Determine the maximum number of loading sequences the component may be loaded with between two inspections. Assume that cracks of length 1 mm may remain in the structure after an inspection. W >> a The material has the fatigue threshold value Kth = 8 MN/m3/2.

Chapter 6

Page 6:21

For crack growth calculations Paris law may be used: da = C (KI)n dN where C = 2.510 12 (units in MN and meter) and n = 3.2. Solution: First investigate if all stress cycles will contribute to the crack propagation. The stress intensity range for the component with the edge crack is (Case 5 in Appendix 3 in the textbook, use that W >> a) a KI = 0 (a) = 1.12 0 a f5 a W The stress range 0 = 92 MPa gives a 0.001 = 5.77 MN/m / 2 (b) = 1.12 92 KI = 0 a f5 W which is below the threshold value Kth = 8 MN/m3/2. It is concluded that at least initially the lower stress range cycles 0 = 92 MPa do not contribute to the crack propagation. Check also for the next stress range cycle: 0 = 200 MPa. One obtains a 0.001 = 12.6 MN/m / 2 KI = 0 = 1.12 200 a f5 W This cycle will induce crack propagation at all crack lengths.

(c)

Determine at which crack length the stress range cycles 0 = 92 MPa will start to contribute to the crack propagation. One has a a1 = 8 MN/m / 2 (d) = 1.12 92 KI = 0 a f5 W giving a1 = 1.92 mm. Thus, as long as the crack is shorter than 1.92 mm, only the cycles with stress range 200 MPa and larger will make that the crack grows. When the crack has reached 1.92 mm all cycles (all four stress ranges) will be damaging to the structure (i.e. give crack propagation). The crack growth must be calculated i two steps: first the growth from 1 mm to 1.92 mm, and then the growth from 1.92 mm to 5 mm. Paris law gives
da = C (KI)n dN (e)

Page 6:22

Chapter 6

Using notation Ns for number of loading sequences, one obtains da n n = C (KI i )n = C (1.12 ) {1 285n + 3 270n + 12 200n } ( a) dNs Separating and integrating give 0.00192 da n = C (1.12 ) {1 285n + 3 270n + 12 200n } Ns 1.6 0.001 a

(f)

(g)

Solving for Ns gives Ns = Ns1 = 2867 sequences, giving N = N1 = 286760 = 172 000 cycles (of these 172 000 cycles 286744 = 126 100 cycles do not contribute to the crack growth). When the crack has reached length 1.92 mm, all stress range cycles contribute to the crack propagation. One obtains da = C (KI i )n dNs
= C (1.12 ) {1 285n + 3 270n + 12 200n + 44 92n } ( a ) Separating and integrating give 0.005 da n = C ( 1.12 ) {1 285n + 3 270n + 12 200n + 44 92n } Ns 1.6 0.00192 a
n n

(h) (i)

Solving for Ns now gives Ns = 2254 sequences (giving N = N2 = 225460 = 135 240 cycles). In total, 2867 + 2254 = 5121 sequences, giving 307 000 cycles may be allowed between two inspections. Paris law has been used for crack length a = 5 mm. The largest stress range cycle is 0 = 285 MPa. This gives the stress intensity range
KI = 1.12 0 0.005 = 40 MN/m3 / 2 a = 1.12 285 (j)

It is concluded that after reduction with a safety factor s (s > 1) the fracture toughness KIc of the material must be at least 40 MN/m3/2. Thus, one must have KIc > s40 MN/m3/2. Alternative solution: Solve the problem by use of the equivalent stress range eq.

Chapter 6

Page 6:23

For crack length less than 1.92 mm, the equivalent stress range becomes
equivalent
{ (0i )n ni } 1 / n 1 2853.2 + 3 2703.2 + 12 2003.2 1 / 3.2 = = ni 16

= 223.85 MPa

(k) (l)

Paris law gives

da n a ) = C (KI)n = C (1.12 equivalent dN

Separating and integrating give 0.00192 da n = C (1.12 223.85 ) N n 0.001 ( a )

(m)

Solving for N gives N = 45 867 cycles. Dividing N by 16 (the number of damaging cycles in a sequence) gives the number of sequences Ns to make the crack grow from length 1 mm to length 1.92 mm. One obtains Ns = Ns1 = 45 867 / 16 = 2867 sequences, which is in agreement with the result obtained above. For crack length larger than 1.92 mm, the equivalent stress range becomes
equivalent
{ (0i )n ni } 1 / n = ni
1 2853.2 + 3 2703.2 + 12 2003.2 + 44 923.2 1 / 3.2 = = 155.13 MPa 60

(n)

Separating and integrating Paris law give 0.005 da n = C ( 1.12 155.13 ) N n 0.00192 ( a )

(o)

Solving for N gives N = 135 247 cycles. Dividing N by 60 (the number of damaging cycles in a sequence) gives the number of sequences Ns to make the crack grow from length 1.92 mm to length 5 mm. One obtains Ns = Ns2 = 135 247 / 60 = 2254 sequences, which is in agreement with the result obtained above. Answer: In total, Ns = 2867 + 2254 = 5121 sequences may be allowed between two inspections, giving 307 000 cycles. For crack length a < 1.92 mm the stress range cycle 0 = 92 MPa will not contribute to the crack propagation.

Page 6:24

Chapter 6

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 7
An introduction to fatigue

No problems in this chapter

FMCHAP7P.DOC/2010-01-29

Chapter 7

Page 7:1

Empty page.

Page 7:2

Chapter 7

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 8
Stress-based fatigue design
Problems with solutions 8/1. A generator is rotating with a speed of 3000 rpm (revolutions per minute). Determine the number of loading cycles N the axle is subjected to due to its own weight. The machine is operating 7000 hours per year during 20 years. Solution: 3000 revolutions per minute give 3000 loading cycles per minute. It gives 3000 60 7000 loading cycles per year, giving 3000 60 7000 20 = 25.2 109 loading cycler per 20 years. Answer: Number N of loading cycles is N = 25.2 109 cycles (thus, it may be exposed to fatigue).

8/2. Estimate the fatigue limit in tension/compression for alternating and pulsating loading of a steel (approximately SS 2225) with the ultimate strength U = 810 MPa. Sketch the Haigh diagram. At an investigation of a similar material the relationships
FL = 0.468 U 50 MPa and FLP 0.85 FL

were found. Use these relations and other thumb rules to estimate the fatigue limits (alternating and pulsating) of the material. Solution: The relationship FL = 0.468U 50 MPa gives, for alternating loading: FL 0.468810 50 MPa = 380 50 MPa, i.e., the fatigue limit FL in tension/compression at alternating loading may, according to this relationship, be expected to fall in the range 330 MPa to 430 MPa. For pulsating loading the relationship between pulsating and alternating
FMCHAP8P.DOC/2010-01-29/TD

Chapter 8

Page 8:1

loading, FLP 0.85FL, gives FLP = 0.85FL = 0.85 (380 50) MPa = 320 42 MPa. Thus, the fatigue limit in tension at pulsating loading is expected to fall in the range from 280 280 MPa (from 320 42 MPa) to 360 360 MPa (from 320 + 42 MPa). Also, compare with other materials. Material data for SIS 2225-03, -04 and -05 gives, for rotating bending, that FLB should be in the range 350 MPa to 460 MPa. The engineering rule FL 0.8FLB gives that FL should fall in the range 280 MPa to 370 MPa. Another engineering rule says that FLR (= FLB) decreases from 0.45U for mild steel to 0.38U for the harder materials. The most unfavourable case, FLR = 0.38U gives FLR = FLB 308 MPa, while the case FLR = 0.45U gives FLR = FLB 365 MPa. This gives, using FL 0.8FLB, that FL should be in the range 250 MPa to 290 MPa. Finally, select, hopefully on the safe side, for alternating loading FL = 280 MPa and for pulsating loading FLP = 0.85FL, which gives FLP = 240 240 MPa. (In practice, fatigue limits will probably be higher than the values estimated here.) Answer: The fatigue limit FL at alternating loading is estimated to 280 MPa, approximately, and the fatigue limit FLP at pulsating loading is estimated to 240 240 MPa, approximately.

8/3. Sketch the Whler (SN, Stress-Number) curve for a normalised 37Mn Si5 steel bar with an ultimate strength of U = 810 MPa. The bar is subjected to a rotating bending moment. At an investigation of a similar material the following fatigue lives were found: FLR 0.45U yields N 106 cycles, FLR 0.90U yields N = 1000 cycles, and FLR = U yields N =1 cycle. Solution and answer: See figure.
1.0 0.9 0.45 0 1 3 6 log N
FLR / U

Page 8:2

Chapter 8

8/4. (a) Draw the fatigue diagram according to Haigh for the normalised steel SS 1650-01 subjected to loading in pure tension/compression. (b) Draw the fatigue diagram according to Goodman for the normalised steel SS 1650-01 subjected to loading in pure tension/compression. Solution:
a (MPa)

200 180
m

(a) Material data gives FL = 200 MPa, FLP = 180 180 MPa, Y = 310 MPa and U = 590 MPa, from which the Haigh diagram can be constructed, see figure.

180 590 360 310 200 180

m

310

590 MPa

+ a (MPa) -

180

310

(b) The same data gives the Goodman diagram according to the second figure. It is noted that this diagram is more complicated to draw than the Haigh diagram (one has both mean value m and amplitude a on the ordinate). It does not, however, contain more information than the Haigh diagram. (From m now on, only the Haigh diagram will be 590 MPa used.) Answer: See figures.

- 200
8/5.

- 310

Constant-amplitude fatigue strengths for materials subjected to different cyclic loading conditions are often expressed in Haigh a B diagrams. In the Haigh diagram given three A straight lines (A, B and C) are shown. Each line C corresponds to constant loading conditions. The 200 loading varies sinusoidally with mean value m, amplitude a, maximum value max (= m + a), 100 m minimum value min (= m a), and stress ratio 0 0 100 200 300 400 MPa R = min /max. Determine which one of these five loading variables is constant in the three cases shown in the figure.
Chapter 8 Page 8:3

Solution: For line A one notices that the sum m + a is a constant (= 300 MPa). Thus, line A gives that stress max is constant. For line B one notices that ratio a / m is constant (= 2). Thus, stress ratio R is constant. For line C one notices that mean stress m = 200 MPa + a, giving m a = 200 MPa = min. Thus, min is constant. Answer: Line A: stress max is constant, line B: stress ratio R is constant, line C: stress min is constant. 8/6. Two axles with the same geometry (but of different size) are machined from the normalised steel SS 1650-01 (U = 590 MPa). The axle diameters are 10 mm and 100 mm, respectively. One axle was machined from a raw material with a diameter 15 mm and the other from a raw material with diameter 120 mm. The loading is rotating bending. Estimate the ratio of the reduced fatigue limits of the two axles if they have the same surface finish. Solution: In Case (a), axle of diameter 10 mm, and in Case (b), axle of diameter 100 mm, the following reduction factors are obtained: due to the size (volume) of the raw material (15 mm and 120 mm, respectively): (a) (a) = 1.0, and (b) (b) 0.8, due to loaded volume: (a) (a) = 1.0, and (b) (b) 0.9, due to surface finish: (a) (a) = , and (b) (b) = , where is unknown, but the same in the two cases. One obtains for Case (a) that (a)(a)(a) = 1.0, and for Case (b) that (b)(b)(b) 0.72 The reduced fatigue limits red FL give
Case (b) red, FL Case (a) red, FL

0.72 FL = 0.72 1.0 FL

Answer: The ratio of the two reduced fatigue limits is, approximately, 0.72.

Page 8:4

Chapter 8

8/7. Assume that the fatigue limits of polished test specimens subjected to alternating loading in tension/compression are given by (ultimate strength is Rm = U)
FL = Rm 2 when Rm is less than 1000 MPa, and Rm 1000 5 when Rm is greater than 1000 MPa.

FL = 500 +

Estimate how the fatigue limits of different materials depend on the ultimate strength. Use some different surface roughnesses, for example a polished surface, a machined surface and a surface with a notch (a standard notch, see Figure 8.7). Solution: Factor for reduction of fatigue limit due to surface finish, and formulae above, give Rm = U = Polished specimen ( = 1.0) FLpol = 250 500 600 MPa 0.68 600 = 410 MPa 0.4 600 = 0.8250 = 200 0.59500 = 295 240 MPa 500 1000 1500 MPa

Machined specimen FLred = FLpol 0.87250 = 220 0.78500 = 390 Notch FLred = FLpol

It is concluded that high-strength steels are very sensitive to surface irregularities (surface finish) and notches. Using the reduction factor Kr, where = 1/Kr, the KTH Handbook, Sundstrm (1998), gives for some values of the surface roughness measure Ra

Chapter 8

Page 8:5

Rm = U = Polished specimen FLpol = Ra = 10 m

500

1000

1500 MPa

250

500

600 MPa 0.60 600 = 360 MPa 0.25 600 = 150 MPa

FLred = FLpol 0.90250 = 225 0.65500 = 325 Ra = 100 m FLred = FLpol 0.62250 = 155 0.33500 = 167

Also here it is noted that the high-strength steels are sensitive to surface irregularities (surface finish) and notches. Answer: Factor for reduction of fatigue limit due to surface finish, and the two formulae given in the problem, give reduced fatigue limits as given in the tables above. It is concluded that high-strength steels are very sensitive to surface roughness and to notches. 8/8. The mean value of the fatigue limit of a welded 4 mm thick aluminium plate has been found to be 92 MPa and the standard deviation of the spread of the fatigue limit measurements is 6 MPa. Determine the stress level at which the probability of fatigue failure is 0.1 per cent. The spread of the fatigue limit is assumed to have a normal (Gaussian) distribution. For the normal distribution function (x) one has (x) = x = 0.50 0 0.90 1.28 0.99 2.33 0.999 3.09 0.9999 3.72

Solution: The mean value of the fatigue limit (here 92 MPa) gives the failure probability 50 per cent. The standard deviation s is s = 6 MPa. The failure probability 0.1 per cent is obtained at the stress level = mean 3.09s = 92 3.096 = 73.5 MPa (where 3.09 was obtained from the table). This implies that in mean one plate out of 1000 plates loaded at the stress level 73.5 MPa is expected to fail due to fatigue.
Page 8:6 Chapter 8

(At stress lever mean 2.33s = 78 MPa the failure probability is one per cent, which means that one plate out of 100 is expected to fail due to fatigue if the plates are loaded at this stress level, and so on.) Answer: At stress level 73.5 MPa the probability of fatigue failure is expected to be 0.1 per cent. 8/9. The fatigue lives of welded beams have been determined at seven (n = 7) identical tests. Stress range 250 MPa was used. The numbers of cycles to failure were 49 000, 61 000, 71 000, 81 000, 88 000, 110 000, and 135 000. Determine the allowable number of cycles at the stress level given, if the probability of failure is not allowed to exceed 1 per cent with confidence C = 0.95. The table below gives the k-factors for different values of n and P and with confidence C = 0.95.
Tolerance limits for the normal distribution (from Rde and Westergren (1998)). The table gives factors k1 and k2 such that the following kind of statements can be made: At least the proportion P is less than x + k1 s with confidence 0.95. At least the proportion P is greater than x k1 s with confidence 0.95. At least the proportion P is between x k2 s and x + k2 s with confidence 0.95. n is sample size, x is sample mean, and s is sample standard deviation. P = 0.90 n 6 7 8 9 10 15 20 25 k1 3.006 2.755 2.582 2.454 2.355 2.068 1.926 1.838 k2 3.733 3.390 3.156 2.986 2.856 2.492 2.319 2.215 P = 0.95 k1 3.707 3.399 3.188 3.031 2.911 2.566 2.396 2.292
Chapter 8

P = 0.99 k2 k1 5.062 4.641 4.353 4.143 3.981 3.520 3.295 3.158 k2 5.758 5.241 4.889 4.633 4.437 3.885 3.621 3.462
Page 8:7

4.422 4.020 3.746 3.546 3.393 2.965 2.760 2.638

Solution: First calculate the logarithm of the number of cycles N to fatigue failure. It gives xi = log Ni = 4.69, 4.79, 4.85, 4.91, 4.94, 5.04, and 5.13 Assume that xi (i.e. log Ni ) has a normal distribution. This means that it is assumed that the xi:s are samples taken from a normally distributed process. Determine the mean value x of xi and the standard deviation s. One obtains
xi x= = 4.9074 n (which gives N = 104.9074 = 80 789 cycles) and s=
n

1 n (xi x )2 = 0.1496 n 1

The probability of fatigue failure should be 1 per cent at confidence level 95 per cent. A table of statistics then gives, using number of samples n = 7, probability P = 0.99, and confidence level C = 0.95: k = 4.641 Allowable value of x at failure probability 0.01 (= 1 P) becomes x = xallowable = x k s = 4.9074 4.641 0.1496 = 4.2131 Number of cycles Nallowable that can be allowed at failure probability 1 per cent is
Nallowable = 10 allowable = 10 4.2131 = 16 334
x

Answer: Allowable number of cycles is, for 1 per cent failure probability, Nallowable = 16 300 cycles.

8/10. A large plate has a small elliptical hole in it. The ratio of the axes of the hole is 2 to 1. Determine the stress concentration factor Kt when the plate is loaded in parallel to the (a) minor axis, (b) major axis. Solution: Handbook (for example Appendix 2 in the textbook) gives, Kt = (1 + 2b / a). Using b / a = 2 and b / a = 1/2 one obtains Kt = 5 and 2, respectively.
Page 8:8 Chapter 8

Comment: A large stress concentration implies a large probability of fatigue failure if the loading varies with time. Consequently, notches and other geometric irregularities should be placed so that the stress concentration is avoided or made as small as possible (i.e., use large radii at the geometric irregularities). Answer: The stress concentration factor Kt is (a) Kt = 5, and (b) Kt = 2. A ground (surface finish Ra = 3 m) axle of steel SS 1650-01 (U = 590 MPa) has been r = 5 mm M M manufactured from a circular bar of 200 mm diameter. The axle has a change of diameter from 180 mm to 160 mm, with the shoulder fillet radius r = 5 mm, see figure. Determine the maximum rotating bending moment M the axle may be subjected to if a safety factor of 2.0 with respect to fatigue failure should be used.

8/11.

Solution: The material has fatigue limit FLB = 270 MPa when loaded with an alternating bending moment. The ultimate strength of the material is U = 590 MPa. The factors reducing the fatigue limit are = 0.80 (diameter 200 mm), = 1 here, because the fatigue notch factor Kf will be used and this factor takes over the role of (else, one would have had = 0.87), and = 0.92 (ground axle with ultimate strength U = 590 MPa). Stress concentration gives Kt = 2.25 (here D / d = 1.125 and r / d = 0.03). The fatigue notch factor Kf becomes Kf = 1 + q (Kt 1) = 1 + 0.84 (2.25 1) = 2.05 where q = 0.84 has been taken from Figure 8.11 in the textbook. Using safety factor sa = 2.0 one obtains FLB which gives nom = 0.18 270 MPa Kf nom allowable = allowable = sa sa Kf FLB Using
nom allowable = 32M = 0.18 270 106 Pa 3 d M = 19 kNm.

one obtains the bending moment

Answer: Maximum allowable bending moment M is M = 19 kNm.

Chapter 8 Page 8:9

8/12. An axle has a change of diameter from 40 mm to 30 mm with a shoulder fillet with radius r = r = 1 mm MT MT 1 mm, see figure. The fillet is ground (R = 5 a 30 40 m). The material is SS 1550-01 with an ultimate strength U = Rm = 490 MPa and a fatigue limit (in torsion) FLT = 140 MPa. Determine the maximum alternating torque MT the axle may be subjected to if a factor of safety 3.0 should be used? Solution: The shear stress at the shoulder fillet is
16MT

d 3 Figures 8.7, 8, 9, and 11, and Appendix 2 in the textbook give = 0.94, = 0.90, = 1, q = 0.70, and Kt = 1.85. The fatigue notch factor becomes Kf = 1 + q (Kt 1) 1 + 0.70 (1.85 1) = 1.6 One obtains 16MT 1 1 1.6 = polished = 0.94 0.90 1 140 106 Pa 3 s 3 d

shoulder = Kf nom = Kf

which, using d = 0.030 m, gives MT 130 Nm. Answer: The maximum allowable torque is MT = 130 Nm. 8/13.
r = 1.5 mm P P

26

30 mm

An axle has a groove (ground, Ra = 7 m) according to the figure. The axle is subjected to a tensile loading P = 18 12 kN. The material is 1450-01 with U = 430 MPa, Y = 250 MPa, FL = 140 MPa and FLP = 130 130 MPa.

Determine the safety factor with respect to fatigue failure if the ratio of the mean value and the amplitude of the loading stress is constant.

Page 8:10

Chapter 8

Solution:
a

Reduce the fatigue data. One obtains

red FL = FL = 0.95 0.95 1 140 = 126 MPa
0.544 1
m

(MPa) 117 MPa

140 126

red FLP = FLP = 0.95 0.95 1 130 = 117 MPa

Draw the Haigh diagram, see figure. 130 180 250 430MPa The notch (the groove) gives the stress concentration factor Kt 2.6. The fatigue notch factor Kf becomes (notch sensitivity factor q from Figure 8.11 in the textbook) Kf = 1 + q (Kt 1) 1 + 0.7 (2.6 1) = 2.12 Stresses in the material at the notch become: mean value m = 2.6 Pm / A and amplitude a = 2.12 Pa / A (the cross-sectional area A is calculated using the smallest diameter, d = 26 mm, at the notch). Ratio a / m becomes
a 2.12 12 = = 0.544 m 2.6 18

A straight line with this slope is entered into the Haigh diagram. The intersection of this line with the reduced fatigue limit of the material gives that the allowable mean stress m is m 180 MPa. Allowable mean value Pm of the tensile force P can now be calculated. One has 4Pm Kt 2 = m = 180 106 Pa, which gives Pm = Pallowable = 36.7 kN d Without safety factor the mean value of the froce can be Pallowable = 36.7 kN. But, according to the problem, the applied load was P = Papplied = 18 kN. The safety factor then is Pallowable 36.7 s= = 2.0 Papplied 18 Answer: The safety factor is s = 2.0, approximately. 8/14. When testing smooth specimens (polished) made from a high strength steel with yield limit Y = 1510 MPa and ultimate strength U = 1800 MPa, fatigue limits at N = 107 cycles were obtained at the following stress levels: 800 400, 400 560, 0 720, 400 860, and 800 850 MPa. Draw the Haigh diagram for notched cylindrical specimens of diameter D = 11 mm. The notch is 1.5 mm deep with a radius in the trough of 0.5 mm and it is polished.
Chapter 8 Page 8:11

Solution: Determine the reduction factors , , , the stress concentration factor Kt, and the fatigue notch factor Kf. Diameter D = 11 mm gives 0.98 1.0. No bending or torsion give = 1.0. Polished surface give = 1.0. Thus = 1.0, which means that no reduction due to surface finish and volume is needed here. Stress concentration: D / d = 11/8 = 1.38 and / d = r / d = 0.5/8 = 0.06 give Kt = 2.8. Fatigue notch factor Kf = 1 + q (Kt 1) = 1 + 0.95 (2.8 1) = 2.7. To make the Haigh diagram valid for the notched specimen, the mean value (corresponding to a static load) should be reduced by the factor Kt and the amplitude by the factor Kf. The point of fatigue limit 800 400 then moves to 800 / 2.8 400 / 2.7 = 286 148 MPa. For the points given, the following fatigue limits are obtained: 286 148, 143 207, 0 267, 143 318, and 286 315 MPa. The yield strength of the material is Y = 1510 MPa. If yielding is to be avoided, the Haigh diagram should be limited at stress Y / Kt = 1510 / 2.8 = 540 MPa both on the abscissa and on the ordinate. Answer: m = m / Kt, a = a / Kf and m + a < Y / Kt, where Kt = 2.8 and Kf = 2.7, which give fatigue limits 286 148; 143 207; 0 267; 143 318; and 286 315 MPa.

8/15. The right end of the cantilever beam (see figure) is moved up and down; the displacement upwards is 2a and downwards a. Determine the maximum allowable value of a with respect to fatigue failure. Use safety factor s = 2.0.
r = 10

65

d = 40

200

The surface of the beam is polished. The r = 5 (mm) material has the modulus of elasticity E = 206 GPa, the yield limit 390 MPa and the ultimate D = 50 strength 590 MPa. Fatigue limits are 270 200 mm MPa at alternating bending and 240 240 MPa at pulsating bending.

Page 8:12

Chapter 8

Solution: The force P that loads the free (right) end of the beam is obtained from PL 3 3EI = which gives P = 3 3EI L Here = 0.5a 1.5a, L = 400 mm and I (= D4 / 64) is the second moment of area of the circular beam cross section with diameter D = 50 mm (implying that here the influence of the notch on the beam deflection is neglected). At the notch the bending moment Mb, notch is obtained as L 3EI 3EI Mb, notch = P = 2 = 2 (0.5 a 1.5 a ) 2 2L 2L At the shoulder the bending moment becomes 3EI 3EI Mb, shoulder = P L = 2 = 2 (0.5 a 1.5 a ) L L Stress concentration at the notch: Ktnotch = 1.8 and q = 0.83 give Kfnotch = 1.66. Stress concentration at the shoulder: Ktshoulder = 1.4 and q = 0.88 give Kfshoulder = 1.35. The stresses at the notch and at the shoulder may now be determined (as a function of deflection a). One obtains Mb, notch 3EI = 2 (Ktnotch 0.5 a Kfnotch 1.5 a ) notch = Wb, notch 2L Wb, notch
and shoulder = Mb, shoulder 3EI = 2 (Ktshoulder 0.5 a Kfshoulder 1.5 a ) Wb, shoulder L Wb, shoulder

where Wb is the section modulus in bending. Reduction factors: = 1, = 0.87, = 1 The reduced fatigue limit at alternating loading a (MPa) becomes 270 210 MPa 235 FLB,red = 0.87270 = 235 MPa and at pulsating loading it becomes m FLBP,red = 240 0.87240 = 240 210 MPa. 590 This gives a Haigh diagram according to the m 240 390 (MPa) figure. The loading is such that the ratio of the amplitude to the mean value is constant. This gives at the notch:
notch a
notch m

1.66 1.5 = 2.77 1.8 0.5

Chapter 8 Page 8:13

Enter a straight line of slope 2.77 in the Haigh diagram. This line intersects the reduced fatigue limit at m 82 MPa. Fatigue failure is thus expected when the mean value of the stress is m = 82 MPa (the stress amplitude is then 2.77m = 227 MPa). The mean value m of the stress at the notch will now be determined from the calculated bending moment. Using the stress concentration factor Ktnotch = 1.8 on the nominal mean stress applied, and using the safety factor s = 2 on the reduced fatigue limit m = 82 MPa determined, one obtains mean Mb, 3EI 1 32 notch 6 1 1.8 nominal = m which gives 1.8 = 1.8 2 0.5 a 3 = 82 10 Pa s 2 Wb, notch 2L d From this a = 0.00048 m = 0.48 mm is solved. At the shoulder is obtained:
shoulder a 1.35 1.5 = = 2.9 1.4 0.5 shoulder m

Enter a straight line of slope 2.9 in the Haigh diagram. The line intersects the reduced fatigue limit at m 78 MPa. Fatigue failure is thus expected when the mean value of the stress is m = 78 MPa (the stress amplitude is then 2.9m = 226 MPa). The mean value m of the stress at the shoulder will now be determined from the calculated bending moment. Using the stress concentration factor Ktshoulder = 1.4 on the nominal stress applied to the beam, and using the safety factor s = 2 on the reduced fatigue limit, one obtains
mean Mb, 3EI 32 1 1 shoulder 1.4 nominal = 1.4 = 78 106 = m = 1.4 2 0.5 a 3 2 s Wb, shoulder L D From this a = 0.00058 m = 0.58 mm is solved.

It is concluded that the notch is the most critical part of the structure. The stresses at the notch will be limiting for the displacement a. Answer: Maximum allowable displacement a at the free beam end will be a = 0.48 mm.

Page 8:14

Chapter 8

8/16.
200 100 15

cross section:

200 9 mm

holes

1m

1m

A beam with cross section HE200B is made of the material 1312-00. The beam is simply supported at its ends and loaded by a force P = P0 P0 at its centre point. Determine, with respect to fatigue failure of the beam, the maximum allowable value of P0. Each flange has two rows of holes (machined, Ra = 20 m) with diameter 8 mm. The holes are drilled 100 mm from each other. Material data: U = 360 MPa, Y = 240 MPa, FL = 110 MPa and FLP = 110 110 MPa.

Solution: The largest bending moment in the beam will appear at the mid-point of the beam, and it is (use total beam length 2l) P l l M= = (P0 P0 ) where l = 1 m 2 2 The beam HE200B has the section modulus Wb = 570 10 6 m3 in bending. If the holes had not been there, the stress 0 in the lower flange of the beam would have been M 0 = Wb Due to the holes, stress concentration will appear. It is assumed that the holes are so widely separated that when calculating the stress concentration factor at one hole, the disturbance of the stress distribution from the other holes may be neglected. Study half of the flange width. This can be done because of symmetry. Using r / a = 4/50 and B / a = 100/50 the stress concentration factor Kt is obtained as Kt 2.8. The fatigue notch factor Kf becomes Kf = 1 + q (Kt 1) 1 + 0.8(2.8 1) = 2.44. This gives the maximum stress at a hole situated where the bending moment is the largest. One obtains B P0 l = (2.8 2.44) B 2r 2Wb where B is half the width of the flange, thus B = 100 mm.

Chapter 8

Page 8:15

(MPa) 94 MPa 2.44 2.8 110 240

m

110 94

Reduction of fatigue limits: The flange thickness is 15 mm, the fatigue notch factor Kf is used, where Kf > 1, and the surface of the hole is machined. This give 0.95, = 1, and 0.90. The reduced fatigue limit at a pulsating load then becomes

red FLP = FLP = 0.95 1 0.90 110 MPa = 94 MPa

The allowable stress amplitude at the hole is 94 MPa (with no safety factor). It gives 0.100 P0 1 B P0 l = 2.44 = 94 10 6 Pa 2.44 6 B 2r 2Wb 0.092 2 570 10 which gives P0 40 kN. The stress amplitude was used as the design stress. The mean stress mean (= 2.894 / 2.44 = 108 MPa) is less than 110 MPa, implying that the straight line giving the service stress intersects the horizontal branch of the fatigue limit curve just slightly to the left of the breaking point at 110 MPa. Thus, it was correct to use the stress amplitude a = 94 MPa as the design stress. Answer: The force P = P0 P0 may be allowed to be, approximately, P = 40 40 kN (no safety factor included here). 8/17.
A 32 H A top view beam cross section 3h A 4H
h H P

side view

The structure in the figure is loaded by a force P = P0 P0. Determine, with respect to fatigue failure at hole A, the maximum allowable value of P0. The hole is machined (Ra = 7 m) and it has a diameter of 20 mm. Material data (for SS 1510-00): U = 600 MPa, Y = 320 MPa, FL = 230 MPa and FLP is unknown (use thumb rule FLP 0.85FL ). Geometrical data: H = 200 mm and h = 20 mm (cf. Chapter 1, Section 1.4.8).

Page 8:16

Chapter 8

Solution: The nominal stress at the hole becomes nom = Mz / I where M = P32H, z = H /2 and 4H H 3 (4H 6h ) (H 2h )3 I= 12 12 One obtains nom = 2125P = 2125(P0 P0) N/m2 (P in N). The stress concentration factor Kt at the hole is Kt = 3.0 (here the case a small hole in a large plate subjected to uniaxial loading is used). This gives the fatigue notch factor Kf = 1 + q(Kt 1) = 1 + 0.87(3.0 1) = 2.74 The factors reducing the fatigue limit due to surface finish and volume become: 0.86, 0.90 and = 1. Pulsating loading implies that the fatigue limit FLP should be used. One obtains FLP = 0.85 FL By use of the reduction factors one obtains FLPred = FLP = 0.85 FL = 0.860.9010.85230 MPa = 151 MPa The amplitude P0 of the load gives the stress amplitude Kfnom = 2.74nom. Let Kfnom = FLPred. It gives 2.742125P0 = 151106, which gives P0 = 26 kN. Answer: Without safety factor, the force P may be P = 26 26 kN.

Damage accumulation, counting of load cycles 8/18. A component of a road vehicle is subjected to a repeated loading sequence. One sequence contains the following loading (i.e. stress) amplitudes (the loading is alternating) a = 200, 180, 150, and 100 MPa The number of loading cycles at each stress level is n = 15, 20, 150, and 3000, respectively The Whler curve of the material is, in this stress range, given by the relation a = 55logN + 430 MPa
Chapter 8 Page 8:17

Determine the damage accumulation D due to one loading sequence, and then determine how many loading sequences the component might resist before fatigue failure. Solution: The Whler curve a = 55 logN + 430 of the material gives the fatigue life N at a given stress amplitude a. ( 430 a ) / 55 One obtains N = 10 At the different stress levels the following damages are obtained a (MPa) 200 180 150 100 Ni 15 199 35 112 123 285 106 ni 15 20 150 3 000 damage ni / Ni 15 / 15 199 20 / 35 112 150 / 123 285 3 000 / 106

Damage D due to one sequence is 15 20 150 3 000 D= = 5.77 10 3 + + + 6 15 199 35 112 123 285 10 The expected number of sequences Ns to fatigue failure is
Ns = 1 / D 173

Answer: Damage D due to one sequence is D = 5.77 10 3, and expected number of sequences Ns to fatigue failure is Ns 173. 8/19. For a welded plate, a number of points on the Whler curve have been determined. At the probability of failure p = 50 per cent and at the stress ratio R = 0 the following points were obtained:
max (MPa) Nf (cycles)

189 1104

182 3104

161 1105

131 3105

107 1106

(a) Use linear regression to fit a straight line in a maxlogN diagram to these points. Estimate by use of linear damage accumulation (the Palmgren-Miner rule) how many sequences (as given below) the plate might resist before fatigue failure is expected. One sequence contains the following stress levels and cycles:

Page 8:18

Chapter 8

max (MPa) n (cycles)

120 100

150 50

180 20

160 40

140 60

(b) According to Jarfall (1977), see reference in textbook, the relationship

k N = constant could give a better result at spectrum loading (i.e. loading with different, possibly random, amplitudes). This relation gives a straight line in a logmaxlogN diagram. Fit a straight line in a logmaxlogN diagram to the data given and estimate by the formula so obtained the number of sequences to fatigue failure.

Solution: (a) Fit a straight line in a maxlogN diagram to the measured data. Let the coordinate y be y = max and the coordinate x = logN. This gives x= y= 4 189 4.47712 182 5 161 5.47712 131 6 107

Fit the straight line y = Ax + B to the points. Linear regression gives (n is number of points) n x y x y y A x A= = 43.034945 and B = = 368.78089 2 2 n n x ( x ) The coefficient of correlation r becomes
r= n x y x y [n x ( x ) ] [n y ( y ) ]
2 2 2 2

= 0.98

which means (r close to

1) that the points fall almost on a straight line.

The Whler curve of the material thus becomes max = 43.03 logN + 368.78 MPa From this is solved:
max N = 10 For the different stress levels the fatigue life and the damage are obtained as

( 368.78

) / 43.03

Chapter 8

Page 8:19

max 120 150 180 160 140

Ni 603 815 121 282 24 360 71 028 207 093

ni 100 50 20 40 60

damage ni / Ni 0.165613610 3 0.412262510 3 0.820997210 3 0.563161710 3 0.289724910 3

The damage D due to one sequence becomes D = (ni / N i) = 2.2517610 3 . Expected number of sequences to fatigue failure is 1 / D = 444. (b) Fit a straight line in a logmaxlogN diagram to the measured data. Let y = logmax and x = logN; one then obtains x= y= 4 2.276 4.47712 2.260 5 2.207 5.47712 2.117 6 2.029

This gives the line logmax = A logN + B, where A = 0.12755 and B = 2.8146 (and the coefficient of correlation r now is r = 0.9723). The Whler curve gives from which is obtained max 120 150 180 160 140 Ni 582893 101353 24270 61108 174077 ni 100 50 20 40 60 damage ni / Ni 0.171558110 3 0.493327410 3 0.824061210 3 0.654582610 3 0.344676010 3
N = 10
( B log max ) / A

The damage D due to one loading sequence becomes D = (ni / N i) = 2.488210 3, which gives the expected number of sequences Ns to fatigue failure: Ns = 1 / D 402. Answer: Number of sequences to fatigue failure becomes (a) 444, and (b) 402, approximately.

Page 8:20

Chapter 8

8/20. A part of a structure is subjected to sequences of vibrations. Each sequence consists of free damped vibrations giving rise to stresses with the stress ratio R = 1, i.e., the mean value of each stress cycle is zero. In the point of the material to be investigated, the stress amplitude of the first cycle in each sequence is 0. The damping of the vibrations is characterized by the logarithmic decrement . The logarithmic decrement is defined as = ln ( i / i+1) where i is the amplitude of the vibration cycle i and i+1 is the amplitude of the cycle i + 1, see figure (a). The Whler curve of the material is, for stress amplitudes a < 0, assumed to be a straight line in a diagram with loga on the ordinate (the y axis) and logN on the abscissa (the x axis), cf. equation (8.2b). The slope of the curve is 1/m in the diagram. The material has no fatigue limit. Determine the number of vibration sequences the structure may be expected to survive. The number of cycles to failure at the stress amplitude 0 would be N0.
0 0

log a log
t

Figure (a)

Figure (b)

log N 0

log N

Solution: The Whler curve is a straight line of slope 1/m in a log alog N diagram. The equation of the curve may be written 1 log a = log N + C (a) m The constant C is determined from the condition that the fatigue life is N0 cycles when the stress amplitude is 0. This gives 1 (b) log 0 = log N0 + C m By inserting C as given by relation (b) into the expression (a), the equation of the Whler curve takes the form a 1 N0 log = log (c) 0 m N The stress amplitude of the different stress cycles following the first stress cycle (with amplitude 0) in each sequence will now be determined together

Chapter 8

Page 8:21

with the fatigue life at each stress amplitude. Let 1 = 0, 2 = 1 = 20, ..., i = i 0, and so on. The definition of the logarithmic decrement yields i i = ln = ln (d) i + 1 i Solve (d) for . It gives = exp( ) (here exp( ) stands for e ). Determine the cyclic life N = Ni at stress level i = i 0. The relation (c) gives i 0 1 N0 log = log (e) 0 m Ni This relation gives Ni = N0 / i m . The accumulated damage D due to one loading sequence becomes 1 1 1 1 m 2m 3m 1 D = + + + = + + + + = (1 + m +2m + 3m + .) (f) N0 N1 N2 N0 N0 N0 N0 N0 Use the series expansion
1 when | x | < 1 1x This gives the damage D of one sequence of vibration as: 1 D= N0 (1 m ) 1 + x + x 2 + x 3 + . = (g)

(h)

Let I be the total number of sequences the structure may be subjected to. The accumulated damage then is I D. Fatigue failure is expected when the accumulated damage becomes unity, thus when I D = 1. This gives 1 (i) 1=I N0 (1 m ) Solve for I, which gives
1 D Using numerical values = 0.05 (giving 0.95) and m = 8 one obtains I = 0.33 N0 I = N0 (1 m ) =

Hence, the stress amplitude 0 alone would have given a cyclic life of N0 loading cycles. The free damped vibration after the first cycle (with amplitude 0) reduces, in this case, the fatigue life of the structure to one third of N0. Answer: The number I of sequences to fatigue failure is expected to be 1 I = = N0 (1 m ) where = e D If = 0.05 and m = 8 one obtains I = 1 / D = 0.33 N0 (solution also given in the Example in Section 8.2.3 in the textbook).
Page 8:22 Chapter 8

8/21. At a random loading of a structure, the following time sequence was recorded (force in kN, see also figure): 4 6 5 7 3 4 2 4 0 4 2 4 3 7 6 8 7 9 6 7 2 6 1 3 0 2 0 2 0 5 2 4 3 7 2 5 3 6 5 7 2 3 1 5 4 (= static level at rest)
9 8 7 6 5 4 3 2 1 0
P (kN)

time

Determine (a) the distribution of the peaks (number of peaks larger than or equal to a certain level), (b) the distribution of the troughs (number of troughs smaller than or equal to a certain level), and (c) the distribution of exceedances (the load spectrum). Solution: See answer. Answer:
9 8 7 6 5 4 3 2 1 0
P (kN)

cumulative distribution of troughs

exceedances

peaks 5 10 15 20 25
number

Chapter 8

Page 8:23

8/22. At a random loading of a structure, the following time sequence was recorded (force in kN, same sequence as in Problem 8/21, see also figure in that problem): 4 6 5 7 3 4 2 4 0 4 2 4 3 7 6 8 7 9 6 7 2 6 1 3 0 2 0 2 0 5 2 4 3 7 2 5 3 6 5 7 2 3 1 5 4 (= static level at rest) Identify loading cycles by use of the rain flow count method. Make a list of those cycles whose amplitude is larger than, or equal to, 1 kN. Solution: This solution is given for the load sequence as it is given in the problem. It should be noted, however, that the solution could be slightly simplified if the sequence is rearranged so that it starts at its largest maximum or its smallest minimum. Here, fore example, the part of the sequence from the starting point at P = 4 kN to the minimum at P = 0 could be cut off and moved to the end of the sequence given. If this is done, the final result will be the same, but the problem of obtaining half cycles that must be matched to each other, see below, will be avoided. Now, let one drop of rain-water start to run (flow) from each maximum value and each minimum value of the loading sequence (the starting point and the final point here form a minimum). A total of 44 drops are needed, see the figure.

9 8 7 6 5 4 3 2 1 0

P (kN)

17 3 2 6 8 10 4 1 15 16 14 20 5 7 13 11 12

9 19 21 31 23 25 27 22 24 26 28 29 35 39 37 38 44 40
time

43

41 33 32 36 30 34 42 18

Page 8:24

Chapter 8

Make a list of the run-ways of all drops. For each drop, write down its number (the drop number), its stating point (load level in kN) and its end point (load level in kN). It gives
Drop No 1 2 3 4 5 6 7 8 9 10 11 12 13 14 starts stops at (kN) at (kN) 4 6 5 7 3 4 2 4 0 4 2 4 3 7 7 5 6 0 4 3 4 2 9 2 4 3 4 6

15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

6 8 7 9 6 7 2 6 1 3 0 2 0 2 0 5

7 7 8 0 7 6 6 2 3 1 2 0 2 0 7 2

31 32 33 34 35 36 37 38 39 40 41 42 43 44

2 4 3 7 2 5 3 6 5 7 2 3 1 5

5 3 4 2 7 3 5 5 6 1 3 2 5 4

Collect pairs of drops so that one pair forms a closed loop. Write down the two drop numbers, the minimum value (in kN) of the closed loop and the maximum value (in kN). One obtains
Drops No mini- maximum mum value value 4 5 7 3 2 0 7) 6 0 4 4 9 10, 11 12, 13 14, 15 16, 17 19, 20 21, 22 23, 24 25, 26 27, 28 30, 31 2 3 6 7 6 2 1 0 0 2 4 4 7 8 7 6 3 2 2 5 32, 33 34, 35 36, 37 38, 39 (40 41, 42 (43 (44 3 2 3 5 7 2 1 5 4 7 5 6 1) 3 5) 4)

(1 2, 3 4, 29 5, 6 7, 8 9, 18

Chapter 8

Page 8:25

Here 20 closed loops have been identified. Due to the fact that the sequence does not starts at the largest maximum or the smallest minimum (see the introduction to the solution above) there will be some single drops left: drops No 1, 40, 43, and 44. Also these drops may, however, be grouped into closed loops in the following way: Divide the running way of drop No 1 into two parts, 1a and 1b. It gives
Drop No starts stops 1a 1b 40 43 44 4 5 7 1 5 5 7 1 5 4

If we think of a continuation of the given sequence as a repetition of the first sequence, then the run-way of the drop No 43 would continue the same way as the drop No 1 has from 5 kN and onwards. The drop No 43 would then start at 1 kN, initially it will run to 5 kN and then further on the next roof to 7 kN. Thus drops 43 and 1b together form half a cycle from 1 kN to 7 kN. This half-cycle may be matched to the drop No 40 to form a full loading cycle (one loop). The drop 1a now stops at 5 kN and it is seen that this part may be matched to the drop No 44. The five one-way drop ways above may thus be combined to give two closed loops. Again, the numbers of the two drops, the minimum value (in kN) of the closed loop and the maximum value (in kN) are written down. One obtains
Drop way minimum maximum No value value 1a, 44 4 43, 1b, 40 1 5 7

Page 8:26

Chapter 8

In total 22 loops (cycles) have been obtained. These cycles may now be used in a fatigue analysis. The ranges (twice the amplitude) of he loading cycles are given in the tables above. Sometimes cycles of a small amplitude do not take part in the fatigue process. If cycles with an amplitude less than 1 kN are neglected, then 12 loading cycles are left to form the basis of a fatigue damage analysis. These are
Drops No 4, 27 7, 8 9, 18 10, 11 21, 22 23, 24 minimum maximum value value 0 2 0 2 2 1 7 4 9 4 6 3 25, 26 28, 29 30, 31 35, 40 36, 37 43, 1b, 34 0 0 2 2 3 1 2 2 5 7 5 7

Answer: Twelve cycles with an amplitude larger than or equal to 1 kN are obtained (and ten cycles with an amplitude less than 1 kN). Minimum and maximum values of the twelve cycles are: (0,7), (2,4), (0,9), (2,4), (2,6), (1,3), (0,2), (0,2), (2,5), (2,7), (3,5), and (1,7).

Chapter 8

Page 8:27

Empty page.

Page 8:28

Chapter 8

Solutions to problems in T Dahlberg and A Ekberg: Failure, Fracture, Fatigue - An Introduction. Studentlitteratur, Lund 2002, ISBN 91-44-02096-1.

Chapter 9
Strain-based fatigue design
Problems with solutions 9/1. A flat bar of material SAE 1045 with a rectangular cross section has a small circular hole at its longitudinal centre axis. The bar is subjected to a remote alternating stress = 300 MPa. (a) Estimate by use of Neubers method and the Morrow equation the number of cycles to fatigue failure. Assume that Kf = 0.9Kt. (b) For the maximum stress and strain at the hole draw the hysteresis loop when = 300 MPa. (c) Determine the fatigue life if = 100 200 MPa. Solution: (Compare with the problem solved in Section 9.3.3 in the textbook.) (a) Number of cycles to fatigue failure If the stress state at the small hole in the flat bar were purely elastic, then the stress concentration factor would have been Kt = 3. The fatigue notch factor Kf is (a) K = 1 + q (K 1)
f t

No information for calculation of the notch sensitivity factor q is given here. Instead, Kf = 0.9Kt = 2.7 was given. Thus, use Kf = 2.7. Due to the high stresses close to the hole, the material at the hole will yield locally. Taking this into consideration, the stress concentration factor K and the strain concentration factor K may be determined from equations (9.9a,b) in the textbook. For low stresses, the second term in the material relation (9.2a,b) can be disregarded. This means that here, where the stresses far away from the hole are relatively low (i.e. mainly within the elastic range), the second term will be small when compared to the first term. Expressions (9.9a,b) in the textbook then become
FMCHAP9P.DOC/2010-01-15/TD

Chapter 9

Page 9:1

K =

max max max and K = = / E

(b,c)

where, as can be seen, only the first term (i.e. Hookes law) in (9.2a,b) has been used in (c). The Neuber hyperbola becomes
Kf2 2 = K K = E (d)

The material data given earlier (see the textbok) for the material SAE 1045 and the nominal stress amplitude = 300 MPa now determine the Neuber hyperbola. The local stress amplitude a and strain amplitude a at the hole are obtained as the intersection point between the Neuber hyperbola and the material stress-strain relation for cyclic loading. One obtains Kf22 2.72 3002 = = = 3.2805 a a E 200 000 (e,f) 1/ 0.18 a a 1/n a a = + = + a E K 200 000 1344 This system of equations gives a = max = 499 MPa and a = max = 0.006574. The stress and strain concentration factors K and K then become
K = max 499 max 0.006574 = = 1.663 and K = = = 4.3828 300 / E 300/ 200 000 (g,h)

(Verify: K K = 1.663 4.383 = 7.29 = 2.72 = Kf2, as it should.) Now the number of cycles to fatigue crack initiation (or fatigue failure) may be determined. According to Morrow, equations (9.4) and (9.7) in the textbook, and by use of the mean stress m = 0, one obtains f m 1227 a = (i) ( 2N )b + f ( 2N )c = ( 2N ) 0.095 + 1.0 ( 2N ) 0.66 E 200 000 Using a = 0.006574, the fatigue life 2N = 4600 reversals to failure is obtained, giving N = 2300 cycles to fatigue failure. (b) Display the hysteresis loop for the material at the hole, when the remote loading stress is = 300 MPa In problem (a) above the stress and strain amplitudes at the hole have been calculated when the remote stress amplitude is = 300 MPa. This implies that when the remote stress has its maximum ( = 300 MPa), the upper end point of the hysteresis loop is situated at = max = 499 MPa and = max =
Page 9:2 Chapter 9

0.006574, see point A in the figure below. When the remote stress has its minimum ( = 300 MPa), the stress and strain at the stress concentration (i.e. at the hole) is = max = 499 MPa and = max = 0.006574. These values give the lower end point of the hysteresis loop, see point B in the figure. Choose point A as starting point of the hysteresis loop, see figure below. A change of the remote stress (the stress far away from the stress concentration) will cause a change of stress and a change of strain at the stress concentration (the hole). The changes and are obtained by use of the Neuber hyperbola (9.13d) and the material relation for changes of stress and strain (9.3a). Using = 300 MPa one obtains 2 2 2 2 = Kf () = 2.7 300 = 3.2805 E 200 000 (j,k) 1 / n 1/ 0.18 = +2 = +2 E 2K 200 000 2 1344 From this system of equations = 0.004670 and = 703 MPa are solved. The lower branch of the hysteresis loop, starting at point A, will pass through the point C. The co-ordinates of point C are obtained from C = A = 499 MPa 703 MPa = 204 MPa (l,m) C = A = 0.006574 0.004670 = 0.001904 For the student: verify that the change of the remote stress = 600 MPa will take this branch of the hysteresis loop to the point B.
499 MPa A

- 0.001912
203 MPa D

We now turn to the upper (i.e. the left) branch of the hysteresis loop. At a change of stress = 300 MPa the upper branch of the hysteresis loop, starting at point B, will pass through point D. The coordinates of point D are obtained from
D = B + = 499 MPa + 703 MPa = 204 MPa D = B + = 0.006574 + 0.004670 (n,o) = 0.001904

- 0.006574
0.006574 C B 0.001912 - 203 MPa

- 499 MPa

By use of these values (and perhaps some more) the hysteresis loop may be drawn, see figure.

Chapter 9

Page 9:3

Comment: It is noticed that when the loop passes the point C the external loading (far away from the hole) will give a nominal stress equal to zero, i.e. the bar has been loaded a number of times up to stress = 300 MPa and then unloaded by the stress = 300 MPa so that the total stress far away from the stress concentration is = 0. Thus, when the bar is unloaded (i.e. when = 0), the residual stress at the hole is C (compressive) and the residual strain is C. The residual stress is compressive because the material has been stretched out (by plastic deformation) at the hole and it was too long when the bar was unloaded. On the other hand, when unloading from the remote stress = 300 MPa to stress = 0, the residual stress at the hole will become D (in tension) and the residual strain is D. In this case the material was compressed at the hole so it was made too short when the bar was unloaded from = 300 MPa to = 0. This explains why the residual stress is in tension at point D when the remote stress is = 0. (The remote stress = 0 indicates that the axial force in the bar is zero. This is the case also at the hole. For equilibrium to be fulfilled at the cross section of the bar at the hole, there must be residual stresses both in tension and in compression. This implies that residual stresses in tension close to the hole will be balanced by residual stresses in compression further away from the hole, and vice versa.)

(c) Fatigue life when = 100 200 MPa The same bar as above (Kf = 2.7) will be investigated, but now the bar is loaded to a nominal (remote) stress = 100 200 MPa. The nominal (remote) stress will now vary between 300 MPa and 100 MPa. This implies that the stress at the hole will vary between A (when = 300 MPa) and a point on the hysteresis loop below C (the stress at the hole is C when the remote stress is = 0). Determine some more points on the hysteresis loop that is obtained when = 100 200 MPa.
499 MPa A

- 0.001912
203 MPa D E 0.006574

Select, once again, the starting point of the loop at point A where A = 499 MPa and A = 0.006574. Determine the point on the loop (the stress at the hole) when = 200 MPa. It gives, as in (j,k) above,

- 0.006574
2 2 2 2 = Kf () = 2.7 200 = 1.458 E 200 000 0.001912 C 1/ 0.18 - 203 MPa = + 2 200 000 2 1344

(p,q)

F B
Page 9:4

- 499 MPa

Chapter 9

This system of equations gives = 519 MPa and = 0.002810. The point E on the hysteresis loop is now obtained from E = A = 499 MPa 519 MPa = 20 MPa (r,s) E = A = 0.006574 0.002810 = 0.003764 Finally, determine the turning point of the new loop (the stress at the hole) when = 400 MPa (the remote stress then is = 100 MPa). It gives, as in (j,k) above, 2 2 ( ) K 2.72 4002 f = = = 5.832 E 200 000 (t,u) 1/ 0.18 = +2 200 000 2 1344 This gives = 829 MPa and = 0.007038. The turning point F on the new hysteresis loop is now obtained from F = A = 499 MPa 829 MPa = 330 MPa
F = A = 0.006574 0.007038 = 0.000464

(v,w)

Using point F as starting point, some points on the upper branch of the hysteresis loop may be determined from the stress changes already calculated. This is left to the student as an exercise. The number of loading cycles to crack initiation is obtained, according to Morrow, from f m a = (x) ( 2N )b + f ( 2N )c E The mean value m of the stress becomes m = (A + F) / 2 = (499 330) / 2 = 84.5 MPa, and the strain amplitude will be a = (A F) / 2 = (0.006574 + 0.000464) / 2 = 0.003519. By use of these values in (x), and by use of the parameter values given, one obtains 1227 84.5 a = 0.003519 = (y) ( 2N ) 0.095 + 1.0 ( 2N ) 0.66 200 000 From this 2N = 23 070 reversals, giving N = 11 500 cycles, to fatigue failure is solved. (This result may be compared with the result obtained in Section 9.3.3 for the same structure. There the remote stress = 200 200 MPa gave the fatigue life N = 6 500 cycles to failure.)

Chapter 9

Page 9:5

Answer: (Compare with the problem solved in Section 9.3.3 in the textbook.) (a) At remote stress = 300 MPa the number of cycles N to fatigue failure is N = 2300, approximately, (b) stress and strain at the hole are (end points of the hysteresis loop) a = 499 MPa and a = 0.006574, respectively, and (c) fatigue life is expected to be N = 11 500 cycles to failure (according to Morrow) at remote stress = 100 200 MPa.

9/2. In a structure a notch with stress concentration factor Kt = 2.8 has been found. Assume that the material follows the cyclic stress-strain curve a a 1/n 1/n and for changes: a = + = +2 E K E 2K where E = 210 GPa, K = 1200 MPa and n = 0.19. The structure is loaded so that the nominal stress some distance away from the notch varies between 50 MPa and + 250 MPa. (In addition to this stress, the stress concentration will be added at the notch.) Determine the expected number of load cycles to fatigue failure. Use fatigue notch factor Kf = Kt, the reduction of the cross-sectional area at rupture is = 65 per cent, and the ultimate strength of the material is U = 470 MPa. Solution: The material relation and the Neuber hyperbola give for stress and strain ranges (stress range is = 250 ( 50) = 300 MPa). As we are going to use the Coffin-Manson rule, only ranges are of interest here. The influence of the mean values is disregarded. One obtains
1/n 1/0.19 +2 = +2 = E 2 K 210 000 2 1200 (Kf )2 (2.8 300)2 Neuber: = = = 3.36 E 210 000 This system of equations gives = 650 MPa and = 0.005169. (a) (b)

Coffin-Mansons fatigue law reads U 1 1 = ln = 1.0498 = 3.5 N 0.12 + D 0.6N 0.6 where D = ln E 1 1 0.65
Page 9:6 Chapter 9

(c)

Numerical values give

470 N 0.12 + 1.04980.6 N 0.6 210 000 from which N = 19 200 cycles (approximately) is solved. Thus, failure is expected after, approximately, 19 000 cycles. 0.005169 = 3.5 (d)

Answer: Fatigue failure is expected after N = 19 000 cycles (approximately).

9/3. A test specimen is subjected to a remote nominal stress = 500 200 MPa. The specimen has a notch with stress concentration factor Kt and fatigue notch factor Kf such that Kt = Kf = 2.8. Assume that the material follows the cyclic stress-strain relationships (for amplitudes and changes, respectively) a a 1/n 1/n a = + and = +2 E K E 2K Determine the stress (mean value and amplitude) and the strain (mean value and amplitude) in the material at the place of stress and strain concentration. Then use the calculated values in a low-cycle fatigue analysis to determine the expected fatigue life N of the test specimen. The material has the following cyclic properties: E = 206 GPa, K = 1750 MPa, n = 0.11, Y = 850 MPa, f = 1400 MPa, f = 0.60, b = 0.10, and c = 0.55. Solution: Determine the hysteresis loop obtained when the remote stress is varying between 700 MPa and 300 MPa. Comment: When = 700 MPa, the stress at the stress concentration should be more than 700 MPa, of course, and less than 2.8700 = 1960 MPa, which is the stress one should have had if the material were fully elastic. The upper end point of the hysteresis loop is obtained from the intersection of the Neuber hyperbola and the cyclic stress-strain relation. One obtains Kf2 2 2.82 7002 a a = = = 18.65 (a) E 206 000
a a 1/n a a 1/0.11 + a = + = E K 206 000 1750 (b)

Chapter 9

Page 9:7

From these two equations the upper end point (the turning point) of the hysteresis loop will be found. One obtains max = 1077 MPa and max = 0.017319 (c,d) The stress concentration factor K and the strain concentration factor K may now be determined (if desired). One obtains max max 1077 0.017319 K = = = 1.5386 and K = = = 5.096 (e,f) 700 / E 700 / 206 000 One finds that KK = 7.8418, and that Kf2 = 2.82 = 7.84, as it should. A change = 400 MPa (i.e. twice the amplitude 200 MPa) of the remote stress causes changes of and at the stress concentration. The changes and at the stress concentration are obtained from the intersection of the Neuber hyperbola and the stress-strain relation for changes. One obtains Kf2 ()2 2.82 4002 = = = 6.119 (g) E 206 000
1 / n 1 / 0.11 = = +2 +2 E 2K 206 000 2 1750 This gives the changes of stress and strain at the stress concentration as = 1116 MPa and = 0.005479 (h) (i,j)

Comment: One notices that 2.8 400 = 1120 MPa, which is the change one would have obtained if the material were fully elastic. Here we have got the change = 1116 MPa in (i), which implies that almost all deformation at the notch is elastic at this change of the remote stress. The second term on the right hand side of (h) is thus, in this case, much smaller than the first term. The reason why we have such a large range for the elastic deformation here is that the material is yielding in tension when the change of load is applied, and from that starting point the elastic range is twice the yield limit of the material until yielding starts at compression. The lower end point of the hysteresis loop may now be determined. One obtains min = max = 1077 1116 MPa = 39 MPa and (k)
min = max = 0.017319 0.005479 = 0.01184 (l)

The mean value and the amplitude of the stress and the strain at the notch can now be calculated. One obtains

Page 9:8

Chapter 9

mean = a =

max + min 1077 39 MPA = = 519 MPa 2 2 max min 1077 + 39 MPA = = 558 MPa 2 2 max + min 0.017319 + 0.01184 = = 0.01450 2 2 max min 0.017319 0.01184 = = 0.0027395 2 2

(m) (n) (o) (p)

mean = a =

The fatigue life N is obtained from the Morrow expression: f mean a = (2N ) b + f(2N ) c E which gives 1400 519 0.0027395 = (2N ) 0.10 + 0.60(2N ) 0.55 206 000 From this, N = 33 400 cycles is solved. Thus, the expected number of cycles to fatigue failure is N = 33 400 cycles. Answer: Failure is expected after N = 33 400 cycles (approximately).

(q) (r)

9/4. A structure has a notch with the fatigue notch factor Kf. The structure is designed for cyclic loading, and (maximum) 10 000 loading cycles is expected during the life of the structure. The remote loading (the loading far away from the notch) is = 250 MPa. Determine the highest value of the fatigue notch factor Kf that can be allowed. Material data: The material can be considered linear elastic, ideally plastic with modulus of elasticity E = 210 GPa and yield strength Y = 350 MPa. The ultimate strength of the material is U = 380 MPa and the ductility D = 1.05. Solution: Use the Coffin-Manson rule to calculate the maximum strain amplitude allowed. It gives U a = 1.75 N 0.12 + 0.5 D 0.6 N 0.6 (a) E Numerical values, as given above, give
Chapter 9 Page 9:9

a = 1.75

380 10 000 0.12 + 0.5 1.05 0.6 10 000 0.6 210000

= 0.0030983 (b) The stress at the notch can not be higher than Y = 350 MPa. (In fact, it could be slightly higher than 350 MPa, as the ultimate strength of the material is 380 MPa. Thus, some deformation hardening could take place in the material, but this is not taken into account here.) The Neuber hyperbola Kf2 2 = E (c)

gives, with max = Y = 350 MPa,

Kf2 2502 350 0.003098 = (d) 210 000 From this the maximum value of the fatige notch factor is solved. One obtains Kf = 1.91 (e)

Answer: Maximum allowable value of the fatige notch factor is Kf = 1.91.

9/5. A structure has a notch with the fatigue notch factor Kf. The structure is designed for cyclic loading, and (maximum) 10 000 loading cycles is expected during the life of the structure. The remote loading (the loading far away from the notch) is = 200 MPa. Determine the highest value of the fatigue notch factor Kf that can be allowed.
stress
Y U

E 0.001667

strain 0.005

Material data: The material behaviour at cyclic loading may be regarded as linear elastic, deformation hardening (see figure). The modulus of elasticity is E = 210 GPa, the yield strength Y = 350 MPa, the ultimate strength of the material is U = 380 MPa, at which stress the strain (at failure) is 0.005. The ductility is D = 1.05.

Solution: Use the Coffin-Manson rule to calculate the maximum strain amplitude allowed. It gives U a = 1.75 N 0.12 + 0.5 D 0.6 N 0.6 E
Page 9:10 Chapter 9

(a)

Numerical values, as given above, give 380 a = 1.75 10 000 0.12 + 0.5 1.05 0.6 10 000 0.6 210000
= 0.0030983 (b) This strain is larger than the strain 0.001667 (the elastic limit, see figure), and less than 0.005, so it is concluded that the Neuber hyperbola must intersect the strain-hardening branch of the stress-strain curve given. The equation of this branch is = 335 + 9000 (stress in MPa) Enter the strain = 0.003098 into this equation. It gives stress = 363 MPa. Thus, the maximum stress = 363 MPa can be allowed at the notch.

The Neuber hyperbola gives, with max = 363 MPa,

Kf2 2 = E

(c)

Kf2 2002 (d) 363 0.003098 = 210 000 From this the maximum allowable value of the fatige notch factor is solved. One obtains Kf = 2.43 (e)

Answer: The largest fatigue notch factor that can be allowed is Kf = 2.43. 9/6. A flat bar of a linear elastic, ideally plastic material with a rectangular cross section has a small circular hole at its centre axis. The stress concentration factor at the hole is Kt = 3.0 and the fatigue notch factor is Kf = 2.8. The bar is subjected to an alternating stress = 200 MPa. Use the Neuber method to estimate the stress and the stain at the hole, and then use the Morrow equation to estimate the number of cycles to fatigue failure. Note that the material is assumed to be ideally plastic also at cyclic loading, i.e. no deformation hardening or softening is present. The modulus of elasticity is E = 200 GPa and the yield strength is Y = 400 MPa. Further, f= 1200 MPa, f= 1.0, b = 0.1 and c = 0.62.
Chapter 9 Page 9:11

stress
Y

strain

Solution: Due to the high stresses at the hole, the material will yield locally. The stress concentration factor K and the strain concentration factor K may be written max max max K = and K = = (a,b) / E (Hookes law is valid for stresses below 400 MPa, thus = / E ) The Neuber hyperbola becomes Kf22 = E
(c)

The material data given and the nominal stress amplitude = 200 MPa now determine the Neuber hyperbola. The local (maximum) stress and strain amplitudes a and a, respectively, at the hole are obtained as the intersection of the Neuber hyperbola with the material stress-strain relation. One obtains K 22 2.82 2002 a a = f = = 1.568 (d,e) E 200 000 a is unknown, but a = Y = 400 MPa This gives a = max = 400 MPa and, from (d), a = max = 0.00392. Now the number of cycles to fatigue failure may be determined. Using the mean stress m = 0, Morrows formula gives
f m 1200 (f) ( 2N )b + f ( 2N )c = ( 2N ) 0.1 + 1.0 ( 2N ) 0.62 E 200 000 Using a = 0.00392, the fatigue life N = 13800 (or 13827) cycles is obtained. a =

Answer: The fatigue life N = 13 800 cycles is expected.

9/7. A test specimen with a notch (stress concentration factor Kt = 3.0 and fatigue notch factor Kf = 2.5) is subjected to a remote stress = 300 MPa. The material is assumed to be linear elastic, ideally plastic (also at cyclic loading) with modulus of elasticity E = 210 GPa, and yield limit Y = 650 MPa, see figure.
stress
Y

strain

Determine the expected number of loading cycles to fatigue failure of the specimen. Data: Cross-section reduction at fracture is = 65 per cent and fracture strength is U = 650 MPa.
Chapter 9

Page 9:12

Solution: The Neuber hyperbola gives (for amplitudes) (Kf )2 a a = (a) where = 300 MPa E As the material is ideally plastic, it is concluded that the stress cannot exceed Y = 650 MPa at the point where the stress concentration appears. Using this (i.e. a = Y = 650 MPa) in equation (a), one obtains 2 1 (Kf ) 1 (2.5 300)2 = = 0.00412 (b) a = a E 650 210 000 The Coffin-Manson relationship gives, with a = 0.00412, 650 N 0.12 + D 0.6N 0.6 = 2 a = 2 0.00412 = 3.5 210 000 Where D = ln[1/(1 )] = 1.04982. Solving (c) for N gives N = 8330 cycles. Answer: Fatigue failure is expected after 8300 cycles, approximately.

(c)

9/8. A flat bar of a linear elastic, deformation hardening plastic material has a rectangular cross section. A small circular hole has been drilled in the bar at its centre axis. The stress concentration factor at the hole is Kt = 3.0 and the fatigue notch factor is Kf = 2.8. The bar is subjected to an alternating remote stress = 200 MPa. Estimate the number of cycles to fatigue failure by use of Neubers method and Morrows equation. Note that the material is assumed to be linearly elastic, linearly deformation k = E /10 hardening also at cyclic loading as shown in the figure. Use modulus of elasticity E = 200 GPa, strain slope k = E /10, and the yield strength = 400 Y MPa. Further, f= 1200 MPa, f= 1.0, b = 0.1, and c = 0.62.

stress
Y

Chapter 9

Page 9:13

Solution: Due to the high stresses near the hole, the material will yield locally. The stress concentration factor K and the strain concentration factor K may be written
K = max and K = max max = / E (a,b)

(Hookes law is valid for stresses below 400 MPa; thus = / E in (b).) The Neuber hyperbola becomes Kf22 = E

(c)

The material data given and the nominal stress amplitude = 200 MPa now determine the Neuber hyperbola. The local stress and strain amplitudes a and a, respectively, at the hole are obtained as the intersection of the Neuber hyperbola and the material stress-strain relationship given in the figure. One obtains Kf22 2.82 2002 a a = = = 1.568 (d,e) E 200 000 a = 360 + k a for a > Y = 400 MPa This gives a = max = 432.5 MPa and, from (d), a = max = 0.003625. Now the number of cycles to fatigue failure may be determined. According to Morrow, and by use of mean stress m = 0, one obtains f m 1200 (f) ( 2N )b + f ( 2N )c = ( 2N ) 0.1 + 1.0 ( 2N ) 0.62 a = E 200 000 Using a = 0.003625, the fatigue life N = 17575 (or 17600) cycles is obtained. Answer: The fatigue life N = 17 600 cycles (approximately) is expected.

9/9. A test specimen is loaded with a repeated stress sequence with a nominal stress according to the figure. The test specimen contains a notch with stress concentration factor Kt = 2.7 and fatigue notch factor Kf = 2.5. Assume that the material follows the cyclic stress-strain curve a a 1/n 1/n and at changes: = a = + +2 E K E 2K

Page 9:14

Chapter 9

(a) Determine the stresses and strains at the notch needed for a fatigue life analysis (i.e. determine the mean values of the stress cycles and the amplitudes of the strain cycles at the notch). Use the rain-flow count method for the cycle counting. (b) Determine the expected fatigue life (the expected number of loading sequences before failure) of the test specimen.
Stress (MPa)

600 400 200 0

time

The material has the following cyclic properties: E = 206 GPa, K = 1750 MPa, n = 0.11, Y = 850 MPa, f = 1500 MPa, f = 0.60, b = 0.10, and c = 0.55.

Solution: (a) Stresses and strains at the notch The mean value and the amplitude of the stress and strain at the notch are obtained from the turning points of the hysteresis loops. Determine these points. The stress and strain at the notch, when = 600 MPa, are obtained from the relationships (the Neuber hyperbola and the cyclic material relation for amplitudes): 2 2 K 2( )2 = f = 2.5 600 = 10.922 a a E 206 000 (a,b) 1 / n 1 / 0.11 a a a a = + = + a E K 206 000 1750 It is assumed that after a number of loading sequences, the stress and the strain stabilize as given by the Ramberg-Osgoods stress-strain relation even if the loading is not alternating here. Therefore, the remote stress 600 MPa is used as an amplitude when the upper turning point of the hysteresis loop is determined. From (a,b) the upper turning point of the hysteresis loop is found. This point is marked (1) in the figure to the right below. One obtains 1 = 999 MPa and 1 = 0.010943 Thus, each time the remote stress reaches the level 600 MPa (points (A) and (C) in the figure to the left), the stress and the strain at the notch become 1 = 999 MPa and 1 = 0.010943, respectively (point (1) in the figure to the right).
Chapter 9 Page 9:15

Stress (MPa)

600 400 200 0

Stress (MPa) (A) (C)

(A)

1000 500 0
time

(1) (3) (5)

(E) (E) (B) (D) (D) (D)

(2) -500 (4)

0.010

Remote (nominal) stress giving stress concentration at the notch

Stress and strain at the notch

A change = 400 MPa of the remote stress, from level (A) to level (B) in the figure, gives 2 2 ( ) K 2.52 4002 f = = = 4.854 E 206 000 1 / 0.11 1 / n = + 2 = +2 E 2 K 206 000 2 1750 This gives the stress and strain changes at the notch: = 998 MPa and = 0.004865 The stress and the stain at the notch, when = 200 MPa at (B), become 2 = 1 = 999 998 MPa = 1 MPa and 2 = 1 = 0.010943 0.004865 = 0.006078 This point has been marked as point (2) in the stress-strain diagram. After this the loading stress turns upwards the amount = 400 MPa up to stress level = 600 MPa, see point (C). The stress at the notch then increases = 998 MPa from 2 to 3 = 1 MPa + 998 MPa = 999 MPa = 1, and the strain increases from 2 to 3 = 0.006078 0.004865 = 0.010943 = 1, giving point (3) in the diagram. Point (3) coincides with point (1). A stress change = 600 MPa of the remote stress, from level (C) to level (D), gives

(c,d)

Page 9:16

Chapter 9

2 2 2 2 = Kf () = 2.5 600 = 10.922 E 206 000 1 / n 1 / 0.11 = + 2 = +2 E 2 K 206 000 2 1750

(e,f)

This gives the stress and strain changes at the notch: = 1438 MPa and = 0.007596 The stress and the strain at the notch when = 0 ( is situated at point (D)) become (see point (4) in the figure) 4 = 3 = 999 1438 MPa = 439 MPa and 4 = 3 = 0.010943 0.007596 = 0.003347 After this two loops with stress changes = 400 MPa follow. At next maximum, i.e. when has come to level (E), the stress and the strain at the notch become (see point (5) in the figure) 5 = 4 + = 439 + 998 MPa = 559 MPa and 5 = 4 + = 0.003347 + 0.004865 = 0.008212 The final stress change, from level (D) back to (A), gives the branch from (4) back to (1) in the hysteresis loop. To summarize, these values give the following stresses and strains at the notch (not all these values were asked for in the problem): No of cycles
1

min
1

max
999 999 559

mean
500 280 60

ampl

min

max

mean

ampl

499 0.006078 0.010943 0.008510 0.002433 719 0.003347 0.010943 0.007145 0.003798 499 0.003347 0.008212 0.005780 0.002433

1 439 2 439

Thus, at the notch the following is obtained: - one cycle between the points (1) and (2) in the stress-strain diagram, corresponding to cycle (A)-(B)-(C) for the remote stress, - one cycle between the points (3) and (4) in the figure, corresponding to cycle (C)-(D)-(A) for the remote stress, and - two cycles between the points (4) and (5) in the figure, corresponding to cycles (D)-(E)-(D)-(E)-(D) for the remote stress.
Chapter 9 Page 9:17

(b) Fatige life The fatigue life Ni at each strain amplitude, with the corresponding mean value of the stress, is obtained from the Morrow relationship f mi ai = (g) (2 Ni )b + f (2 Ni )c E which gives
0.002433 = 0.003798 = 0.002433 = 1500 500 (2 N1) 0.10 + 0.60 (2 N1) 0.55 206 000 1500 280 (2 N2) 0.10 + 0.60 (2 N2) 0.55 206 000 1500 60 (2 N3) 0.10 + 0.60 (2 N3) 0.55 206 000 giving giving giving N1 = 63 650 N2 = 20 355 N3 = 195 400

The accumulated damage from one sequence is (use the Palmgren-Miner damage accumulation rule) 1 2 1 1 + + = (h) D= 63 650 20 355 195 400 13 320 From this, S = 1/D = 13 320 sequences to fatigue failure is obtained. Answer: One expects S = 1/D = 13 300 sequences to fatigue failure.

Page 9:18

Chapter 9

Extra problem, a version of 9/1. 9/1x. A flat bar of material SAE 1045 with a rectangular cross section has a small circular hole at its centre axis. The bar is subjected to an alternating stress = 300 MPa. (a) Estimate the number of cycles to fatigue failure by use of Neubers method. Assume that Kf = 0.9Kt. (b) Draw the hysteresis loop for = 300 MPa. (c) Determine the fatigue life if = 150 150 MPa. Solution: (a) Number of cycles to fatigue crack initiation If the stress state at the small hole in the flat bar was purely elastic, then the stress concentration factor would have been Kt = 3. The fatigue notch factor Kf is
Kf = 1 + q (Kt 1) (a)

No information for calculation of the notch sensitivity factor q is given here. Instead, Kf = 0.9Kt = 2.7 was given. Thus, use Kf = 2.7. Due to the high stresses close to the hole, the material will yield locally. Taking this into consideration, the stress concentration factor K and the strain concentration factor K may be determined from equations (9.9a,b). If the second term in the material relation (9.2a,b) is disregarded (for stresses far away from the hole the second term is supposed to be small as compared with the first term) one obtains (9.9a,b) as max max max K = and K = = (b,c) / E The Neuber hyperbola becomes
Kf22 = E (d)

The material data given earlier for the material SAE 1045 and the nominal stress amplitude = 300 MPa now determine the Neuber hyperbola. The local stress amplitude a and strain amplitude a at the hole are obtained as the

Chapter 9

Page 9:19

intersection point between the Neuber hyperbola and the material stress-strain relation (for cyclic loading). One obtains Kf2 2 2.72 3002 = = = 3.2805 a a E 200 000 (e,f) 1/ 0.18 a a 1/n a a = + = + a E K 200 000 1344 This system of equations gives a = max = 499 MPa and a = max = 0.006574. The stress and strain concentration factors K and K then become
K = max 499 max 0.006574 = = 1.663 and K = = = 4.3828 300 / E 300/ 200 000 (g,h)

(Verify: K K = 1.663 4.383 = 7.29 = 2.72 = Kf2, as it should.) Now the number of cycles to fatigue failure may be determined. According to Morrow, (9.4) and (9.7), and by use of mean stress m = 0, one obtains f m 1227 a = (j) ( 2N )b + f ( 2N )c = ( 2N ) 0.095 + 1.0 ( 2N ) 0.66 E 200 000 Using a = 0.006574, the fatigue life 2N = 4600 reversals to failure is obtained, giving N = 2300 cycles to fatigue failure. (b) Display the hysteresis loop for the material at the notch, when the remote loading is = 300 MPa According to problem (a) above the upper turning point of the hysteresis loop is situated at = max = 499 MPa and = max = 0.006574, see point A in the figure below. The lower turning point is situated at = max = 499 MPa and = max = 0.006574, see point B in the figure. Choose the point A as starting point of the hysteresis loop, see figure below. A nominal change of stress (far away from the stress concentration) will cause a change of stress and a change of strain at the stress concentration. The changes and are obtained by use of the Neuber hyperbola and the material relation for changes of stress and strain. Using = 300 MPa one obtains 2 2 2 2 = Kf () = 2.7 300 = 3.2805 E 200 000 (k,l) 1 / n 1/ 0.18 = +2 = +2 E 2K 200 000 2 1344
Page 9:20 Chapter 9

From this system of equations = 0.004662 and = 702 MPa are solved. The lower branch of the hysteresis loop, starting at point A, will pass through the point C. The co-ordinates of point C are obtained from C = A = 499 MPa 702 MPa = 203 MPa (m,n) C = A = 0.006574 0.004662 = 0.001912 Verify that the change of stress = 600 MPa will take this branch of the hysteresis loop to the point B.
499 MPa - 0.001912 203 MPa A

At a change of stress = 300 MPa the upper branch of the hysteresis loop, starting at point B, will pass through point D. The coordinates of point D are obtained from
D = B + = 499 MPa + 702 MPa = 203 MPa D = B + = 0.006574 + 0.004662 = 0.001912 (o,p)

- 0.006574
0.006574 C B 0.001912 - 203 MPa

- 499 MPa

By use of these values (and perhaps some more) the hysteresis loop may now be drawn, see the figure.

Comment It is noticed that when the loop passes the point C the external loading (far away from the hole) will give a nominal stress equal to zero, i.e. the bar has been loaded (a number of times) up to stress = 300 MPa and then unloaded by the stress = 300 MPa so that the total stress far away from the stress concentration will be = 0. At the stress concentration, however, the residual stress C and a residual strain C will remain. (c) Fatigue life when = 150 150 MPa The same bar as above (Kf = 2.7) will be investigated, but now the bar is loaded to a nominal stress = 150 150 MPa. The nominal (remote) stress will now vary between 0 and 300 MPa. This implies that the stress at the hole will vary between C and A according to problem (b).

Chapter 9

Page 9:21

As an extra exercise the hysteresis loop between A A and C will be drawn. Select, once again, the starting point of the loop at point A where A = - 0.001912 499 MPa and A = 0.006574. Determine one 203 MPa D more point on the loop. Here = 200 MPa - 0.006574 will be chosen. It gives, as in (j,k) above, E 0.006574 2 2 2 2 = Kf () = 2.7 200 = 1.458 E 200 000 (q,r) 0.001912 1/ 0.18 C F - 203 MPa = +2 200 000 2 1344
499 MPa B

- 499 MPa

This gives = 519 MPa and = 0.002810. The point E on the hysteresis loop is now obtained from E = A = 499 MPa 519 MPa = 20 MPa (s,t) E = A = 0.006574 0.002810 = 0.003764 The point F on the loop (branch starting at C, see the figure) gets the coordinates F = C + = 203 MPa + 519 MPa = 316 MPa (u,v) F = C + = 0.001912 + 0.002810 = 0.004722 The number of loading cycles to fatigue failure is obtained, according to Morrow, from f m (w) ( 2N )b + f ( 2N )c a = E The mean value m of the stress becomes m = (A + C) / 2 = (499 203) / 2 = 148 MPa, and the strain amplitude will be a = (A C) / 2 = (0.006574 0.001912) / 2 = 0.002331. By use of these values in (v), and by use of the given parameter values, one obtains 1227 148 a = 0.002331 = (x) ( 2N ) 0.095 + 1.0 ( 2N ) 0.66 200 000 From this 2N = 95 600 reversals, giving N = 47 800 cycles, to fatigue failure is solved. (This result may be compared with the result N = 6 500 cycles to fatigue failure obtained when = 200 200 MPa, obtained in Section 9.3.1.) Answer: (Compare with the solution given in example above) (a) Number of cycles N to crack initiation is 2300, approximately, (b) stress and strain at hole is a = 499 MPa and a = 0.006574, respectively (c) Nf = 47 800 cycles (according to Morrow).
Page 9:22 Chapter 9